Release v3.8.35 (#4743)

* chore(release): open v3.8.35 development cycle

* fix db vacuum scheduler settings (#4726)

Scheduled VACUUM now follows Storage page settings (scheduledVacuum/vacuumHour) as single source of truth; env-flag control path removed. 11/11 vacuum-scheduler tests pass against release/v3.8.35 tip; no orphaned env refs. Integrated into release/v3.8.35.

* fix(tier): noAuth providers count as free; free filter returns empty … (#4753)

noAuth providers now classified free (union of legacy list + NOAUTH_PROVIDERS chat-tier derivation), -free arena_elo alias, and auto/<cat>:free returns an empty pool when no free candidate matches (opt-in legacy fallback via OMNIROUTE_AUTO_FREE_FALLBACK_TO_FULL_POOL). New env var documented in .env.example + ENVIRONMENT.md; CHANGELOG bullet added (maintainer co-author). 46/46 node + 56/56 vitest tests pass on release tip; env-doc-sync, docs-sync, typecheck:core, lint, file-size all green. Integrated into release/v3.8.35.

* refactor(chatCore): extrai 11 helpers de nível superior para 6 leaves puros (#3501) (#4571)

chatCore god-file decomposition (#3501): extract 6 pure leaves (cacheUsageMeta, executorClientHeaders, nonStreamingResponseBody, skillsFormat, streamErrorResult, streamFinalize) from chatCore.ts. Rebased onto release/v3.8.35 tip (resolved single chatCore.ts conflict — removed now-extracted inline buildExecutorClientHeaders). 265/265 chatcore tests, 26/26 new leaf tests, typecheck:core, cycles, file-size all green. Integrated into release/v3.8.35.

* refactor(chatCore): extrai resolveExecutorWithProxy + getExecutionCredentials para leaves (#3501) (#4646)

chatCore #3501: extract resolveExecutorWithProxy + getExecutionCredentials to leaves (executorProxy.ts, executionCredentials.ts). Clean cherry-pick onto release tip post-#4571. 12/12 new leaf tests, typecheck:core, cycles, file-size green. Integrated into release/v3.8.35.

* refactor(chatCore): extrai transforms de mensagens Claude p/ leaf (#3501) (#4708)

chatCore #3501: extract Claude upstream-message transforms to leaf (claudeUpstreamMessages.ts + claudeMessageTypes.ts). Clean cherry-pick post-#4646. 8/8 new leaf tests, typecheck/cycles/file-size green. Integrated into release/v3.8.35.

* refactor(chatCore): extrai persistAttemptLogs para leaf (#3501) (#4717)

chatCore #3501: extract persistAttemptLogs to leaf (attemptLogging.ts). Rebased onto release tip post-#4708 (resolved imports conflict: kept tip's resolveCompressionHeader from compression Phase 3, dropped now-unused logTruncation import moved into the leaf). 288/288 chatcore tests, typecheck/cycles/file-size green. Integrated into release/v3.8.35.

* refactor(chatCore): extrai stageTrace + compressionUsageReceipt para leaves (#3501) (#4721)

chatCore #3501: extract stageTrace + compressionUsageReceipt to leaves. Clean cherry-pick post-#4717. 6/6 new leaf tests, typecheck/cycles/file-size green. Integrated into release/v3.8.35.

* refactor(chatCore): extrai prepareUpstreamBody (1ª sub-fatia do executeProviderRequest, #3501) (#4730)

chatCore #3501: extract prepareUpstreamBody (first sub-slice of executeProviderRequest) to leaf (upstreamBody.ts). Clean cherry-pick post-#4721. 7/7 new leaf tests, full 301/301 chatcore suite, typecheck/cycles/file-size green. Completes the 6-PR chatCore decomposition stack into release/v3.8.35.

* fix(db): make db-backup import size cap configurable (#4719) (#4757)

Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com>

* chore(quality): expand check:release-green to the FULL release-PR gate set (#4758)

The release-green pre-flight (Solution C) previously covered only a subset of the
gates that run exclusively on the release PR (PR→main), so reds still accrued
silently on release/** and surfaced in ~40-min layers at release time (v3.8.34:
3 CI rounds — CodeQL sanitization, then the fail-fast Quality Ratchet revealing
openapi then cyclomatic-complexity one push at a time, plus zizmor/integration).

Now check:release-green reproduces the COMPLETE release-PR gate set and reports
EVERY red in one pass (collected, not fail-fast):

- New DRIFT ratchets (report-only, rebaselined at release, never block):
  cyclomatic complexity, dead-code, type-coverage, compression-budget,
  openapi-coverage, workflow-lint (zizmor), codeql-ratchet.
- New HARD gates (real defects): docs-all (fabricated-docs strict + i18n mirror
  sync) and the integration test suite (gated behind !--quick).

The only release-PR gates it still cannot reproduce locally are GitHub-side CodeQL
semantic analysis and SonarQube/SonarCloud (external services).

The nightly-release-green workflow and /green-prs inherit the expanded coverage
automatically (they invoke this script), so cycle drift is now surfaced
continuously and the release PR is green on its first CI run.

Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com>

* fix(dashboard): add missing onboarding.tiers step title (#4698) (#4755)

Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com>

* feat(compression): Output Styles registry + D0 telemetry (Phase 4A) (#4694)

Phase 4A: Output Styles registry + D0 telemetry. Integrated into release/v3.8.35.

* feat(compression): SLM tier for ultra (Phase 4B) [stacked on #4694] (#4707)

Phase 4B: SLM tier for ultra. Integrated into release/v3.8.35.

* feat(compression): context-budget adaptive compression (Phase 4C) [stacked on #4707] (#4716)

Phase 4C: adaptive context-budget compression. Integrated into release/v3.8.35.

* feat(compression): offline evaluation harness (Phase 4 D1) [stacked on #4716] (#4720)

Phase 4 D1: offline evaluation harness. Integrated into release/v3.8.35.

* fix(sse): deepseek-web folds role:tool results into prompt transcript (#4712) (#4756)

Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com>

* fix(dashboard): remove dead unconditional useLiveRequests call in HomePageClient (#4759, #4745, #4596) (#4761)

Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com>

* fix(dashboard): dedupe provider nodes by id on compatible-provider add (#4746) (#4768)

Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com>

* chore(db): re-export compressionRunTelemetry from localDb to satisfy db-rules (#4775)

Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com>

* docs(security): add canonical STRIDE-based threat model (#4783)

Canonical STRIDE threat model. Integrated into release/v3.8.35.

* test(dashboard): add smoke test for home client dashboard (#4793)

Smoke test guarding the dashboard home client render (regression #4745/#4759). Code fix already landed via #4761; this PR's jsdom smoke test is the net-new regression guard. Integrated into release/v3.8.35.

* fix(combos): auto-promote zeroLatencyOptimizationsEnabled so legacy configs (pre-3.8.33 fallbackCompressionMode="lite") round-trip on the first GUI edit (#4774)

Auto-promote zeroLatencyOptimizationsEnabled + strip v3.8.31-era removed keys so legacy combo configs round-trip through PUT /api/combos/{id} on first GUI edit (closes #4382 followup). Pre-merge: rewrote the now-stale reject test to assert auto-promotion + added passthrough/round-trip regression guards; reconciled combos/page.tsx file-size baseline. Integrated into release/v3.8.35.

* refactor(chatCore): extrai parse + usage-stats não-streaming do executeProviderRequest (#3501) (#4762)

chatCore #3501: extract parseNonStreamingResponseBody + recordNonStreamingUsageStats. Integrated into release/v3.8.35.

* refactor(chatCore): extrai recordContextEditingTelemetryHook (#3501) (#4779)

chatCore #3501: extract recordContextEditingTelemetryHook. Integrated into release/v3.8.35.

* refactor(chatCore): extrai recordCompressionCacheStats (#3501) (#4792)

chatCore #3501: extract recordCompressionCacheStats. Integrated into release/v3.8.35.

* refactor(chatCore): extrai writeCavemanOutputAnalytics (#3501) (#4794)

chatCore #3501: extract writeCavemanOutputAnalytics. Integrated into release/v3.8.35.

* refactor(chatCore): extrai scheduleQuotaShareConsumption (POST-hook não-streaming, #3501) (#4780)

chatCore #3501: extract scheduleQuotaShareConsumption (non-streaming POST-hook). Integrated into release/v3.8.35.

* refactor(chatCore): extrai emitRequestGamificationEvent (helper compartilhado DRY, #3501) (#4776)

chatCore #3501: extract emitRequestGamificationEvent (DRY streaming/non-streaming). Integrated into release/v3.8.35.

* refactor(chatCore): extrai runPluginOnResponseHook (#3501) (#4782)

chatCore #3501: extract runPluginOnResponseHook. Integrated into release/v3.8.35.

* refactor(chatCore): extrai scheduleStreamingQuotaShareConsumption (POST-hook streaming, #3501) (#4784)

chatCore #3501: extract scheduleStreamingQuotaShareConsumption (streaming POST-hook). Integrated into release/v3.8.35.

* refactor(chatCore): extrai recordStreamingUsageStats (analytics de usage streaming, #3501) (#4791)

chatCore #3501: extract recordStreamingUsageStats. Integrated into release/v3.8.35.

* refactor(chatCore): extrai recordStreamingCost (custo por-request streaming, #3501) (#4790)

chatCore #3501: extract recordStreamingCost (per-request streaming cost). Integrated into release/v3.8.35.

* docs(readme): credit ponytail + OmniCompress; restore env-doc-sync release-green (#4799)

README compression credits (ponytail/OmniCompress) + env-doc-sync ignore for eval-only OMNIROUTE_EVAL_CREDENTIALS (restores release-green after #4720). Integrated into release/v3.8.35.

* chore(quality): trim combo-config.test.ts comments under file-size cap (#4774 follow-up) (#4800)

Restore file-size release-green. Integrated into release/v3.8.35.

* feat(api-docs): Redoc-rendered /api/docs + consolidate OpenAPI spec to docs/openapi.yaml (#4781)

Redoc /api/docs + OpenAPI spec consolidated to docs/openapi.yaml (canonical 201-path complete spec; old path → legacy fallback). All refs/gates/tests/CI updated. Integrated into release/v3.8.35.

* docs(compression): declare Phase 4 layers — Output Styles, adaptive dial, per-request control (#4801)

The README compression section listed the 9 input engines but not the Phase 4
layers now in production:
- Output Styles (output-axis steering: terse-prose / less-code / terse-cjk, lite/full/ultra)
- adaptive context-budget dial (reserve-output|percentage|absolute · floor|replace-autotrigger|off)
- per-request x-omniroute-compression precedence + the offline eval harness
Also bumped the highlights range to v3.8.35, expanded the compression feature bullet,
and marked the GUIDE's Phase 4 row Shipped (was 'Planned' — it's merged on v3.8.35).

Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* chore(release): finalize v3.8.35 CHANGELOG + docs reconciliation

- CHANGELOG: complete 3.8.35 section (all 35 commits since v3.8.34,
  contributor attribution: @rdself @megamen32 @KooshaPari @JxnLexn)
- docs(security): align THREAT_MODEL.md refs with real code
  (routeGuard.ts, tokenLimits.ts, /api/monitoring/health) — fabricated-docs gate
- check:fabricated-docs: skip docs/superpowers/specs (dated research reports)
- i18n: sync 3.8.35 section into 41 CHANGELOG mirrors (docs-sync size gate)
- ratchet rebaseline: cyclomatic 1916->1920, eslintWarnings 3907->3912
  (inherited cycle drift; release-finalize diff is docs-only)

* fix(release): resolve inherited base-reds surfaced by v3.8.35 release CI

Cycle base-reds that only run on PR→main (not the PR→release fast-path):

- test(autoCombo): suffixComposition-4517 used node:test in a vitest-only dir
  (#4753) → vitest found no suite. Switch to the vitest API. (Vitest job)
- test(agentSkills): openapiParser fixture wrote docs/reference/openapi.yaml;
  parser reads docs/openapi.yaml since #4781 → point fixture at the new path.
  (Unit/Coverage/Node24/Node26 shard 4)
- test(integration): proxy-pipeline source-scan expected inline streaming-cost
  code that #4790/#3501 extracted to the recordStreamingCost leaf → assert the
  delegation instead. (Integration 1/2)
- fix(chatCore): derive the log trace id from crypto, not Math.random
  (CodeQL js/insecure-randomness — log-correlation id, not a secret).
- test(resilience): circuit-breaker invalid-cooldown fallback asserted t>29000,
  flaking on slow CI where ~1.6s elapsed gave t=28401 → tolerate wall-clock
  drift (t>25000). (Unit 6/8)

* fix(usage): derive pending-request id from crypto, not Math.random

CodeQL js/insecure-randomness (#669): the pending-request id generated in
trackPendingRequest (usageHistory.ts) flows into attempt logging and was flagged
as insecure randomness in a security context. It's a log-correlation id, not a
secret — switch to crypto RNG to clear the alert. Pairs with the chatCore traceId
fix in 37c49781a (same sink).

---------

Co-authored-by: Diego Rodrigues de Sa e Souza <souzamiriamrodrigues790@gmail.com>
Co-authored-by: Randi <55005611+rdself@users.noreply.github.com>
Co-authored-by: Demiurge The Single <megamen932@gmail.com>
Co-authored-by: KooshaPari <42529354+KooshaPari@users.noreply.github.com>
Co-authored-by: Jan Leon <Jan.gaschler@gmail.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

.agent/workflows/git-workflow.md

@@ -1,54 +0,0 @@
----
-description: Git workflow — NEVER commit directly to main. Always use feature branches.
----
-
-# Git Workflow
-
-## ⚠️ CRITICAL RULE: NEVER commit directly to `main`
-
-## Steps
-
-1. **Before starting any work**, create a feature branch from `main`:
-
-   ```bash
-   git checkout main && git pull origin main
-   git checkout -b feature/<feature-name>
-   ```
-
-2. **During development**, commit to the feature branch:
-
-   ```bash
-   git add -A && git commit -m "<type>(<scope>): <description>"
-   ```
-
-3. **Before pushing**, verify the build passes:
-
-   ```bash
-   npm run build
-   ```
-
-4. **When the feature is complete and verified**, push the branch and STOP:
-
-   ```bash
-   git push origin feature/<feature-name>
-   ```
-
-5. **DO NOT** create a PR, merge, or push to `main`. Let the user handle that.
-
-## Branch naming convention
-
-- `feature/<name>` — new features
-- `fix/<name>` — bugfixes
-- `refactor/<name>` — refactoring
-- `docker/<name>` — Docker / infrastructure changes
-- `style/<name>` — UI / CSS changes
-
-## Commit types
-
-- `feat` — new feature
-- `fix` — bugfix
-- `refactor` — code refactoring
-- `style` — UI / CSS changes
-- `docker` — Docker / infrastructure
-- `docs` — documentation
-- `chore` — maintenance

.dockerignore

@@ -9,6 +9,7 @@
 # Dependencies and build output
 node_modules
 .next
+.build
 out
 build
 dist
@@ -30,3 +31,95 @@ npm-debug.log*
 yarn-debug.log*
 yarn-error.log*
 .pnpm-debug.log*
+
+# Test suites
+tests
+test-results
+playwright-report
+blob-report
+
+# Documentation
+# Issue #2348: The Dashboard Docs viewer reads markdown from `/app/docs` at
+# runtime. The previous `docs/*` block hid every file except openapi.yaml,
+# so the in-product help screen failed with ENOENT for every page.
+# We now keep the English markdown tree plus the docs assets imported by MDX
+# during `next build`, while still dropping the bulky translated docs and
+# extra raster diagram sources that account for most of the docs footprint
+# of the ~50 MB docs directory. The Docs viewer reads the default-locale
+# (English) sources at runtime, so translations are not required in the
+# container image.
+docs/i18n/**
+docs/diagrams/**/*.png
+docs/diagrams/**/*.jpg
+docs/diagrams/**/*.jpeg
+docs/diagrams/**/*.gif
+docs/diagrams/**/*.webp
+# Note: `*.md` matches the root only (Go filepath.Match does not cross /),
+# so nested docs/**/*.md is implicitly kept without a re-include rule.
+*.md
+!README.md
+
+# Electron (separate build)
+electron
+
+# VS Code extension (separate project)
+vscode-extension
+
+# Build artifacts
+*.tgz
+*.AppImage
+*.deb
+*.rpm
+
+# Package manager lock (bun)
+bun.lock
+
+# Agent config
+.agents
+.gemini
+
+# Misc
+llm.txt
+images
+clipr
+omnirouteCloud
+omnirouteSite
+
+# Temporary/Scratch Folders
+_*
+
+# CI/CD and Version Control (that are not actual code)
+.github
+.husky
+.omc
+
+# Test Configs and Reports
+playwright.config.ts
+vitest*.ts
+audit-report.json
+sonar-project.properties
+
+# Deployment Configs
+docker-compose*.yml
+fly.toml
+
+# Consistent with .gitignore
+.DS_Store
+.idea/
+.config/
+.data/
+.omnivscodeagent/
+*.sqlite-*
+*.tsbuildinfo
+next-env.d.ts
+security-analysis/
+.analysis/
+antigravity-manager-analysis/
+.sisyphus/
+.plans/
+app.__qa_backup/
+.app-build-backup-*/
+.gitnexus
+.worktrees
+.next-playwright/
+cloud/

.github/CODEOWNERS

@@ -0,0 +1,2 @@
+* @diegosouzapw
+

.github/FUNDING.yml

@@ -0,0 +1,5 @@
+# Funding links for OmniRoute — rendered as the "Sponsor" button on GitHub.
+# Docs: https://docs.github.com/repositories/managing-your-repositorys-settings-and-features/customizing-your-repository/displaying-a-sponsor-button-in-your-repository
+github: diegosouzapw
+# Additional platforms (uncomment and fill in before enabling):
+# custom: ["https://omniroute.online/donate"]

.github/ISSUE_TEMPLATE/bug_report.yml

@@ -0,0 +1,171 @@
+name: Bug Report
+description: Report a bug or unexpected behavior in OmniRoute
+title: "[BUG] "
+labels: ["bug"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Thanks for taking the time to report a bug. Please fill out the sections below so we can reproduce and fix the issue.
+
+  - type: input
+    id: version
+    attributes:
+      label: OmniRoute Version
+      description: "Run `omniroute --version` or check the left sidebar in the dashboard."
+      placeholder: "e.g. 3.7.9"
+    validations:
+      required: true
+
+  - type: dropdown
+    id: install-method
+    attributes:
+      label: Installation Method
+      options:
+        - npm (global)
+        - Docker / Docker Compose
+        - Electron desktop app
+        - Built from source
+    validations:
+      required: true
+
+  - type: dropdown
+    id: os
+    attributes:
+      label: Operating System
+      options:
+        - Windows
+        - macOS
+        - Linux
+    validations:
+      required: true
+
+  - type: input
+    id: os-version
+    attributes:
+      label: OS Version
+      placeholder: "e.g. Windows 11 25H2, macOS 26.5, Ubuntu 26.04"
+    validations:
+      required: false
+
+  - type: input
+    id: node-version
+    attributes:
+      label: Node.js Version
+      description: "Run `node --version`. Skip if using Docker."
+      placeholder: "e.g. 24.15.0"
+    validations:
+      required: false
+
+  - type: input
+    id: provider
+    attributes:
+      label: Provider(s) Involved
+      description: "Which AI provider(s) does this affect?"
+      placeholder: "e.g. Antigravity, OpenRouter, Ollama, Qwen"
+    validations:
+      required: false
+
+  - type: input
+    id: model
+    attributes:
+      label: Model(s) Involved
+      placeholder: "e.g. claude-opus-4-7, gpt-5.5, gemini-3.1-pro"
+    validations:
+      required: false
+
+  - type: input
+    id: client-tool
+    attributes:
+      label: Client Tool
+      description: "Which tool are you using OmniRoute with?"
+      placeholder: "e.g. Claude Code, Cursor, Roo Code, OpenClaw, Gemini CLI, cURL"
+    validations:
+      required: false
+
+  - type: textarea
+    id: description
+    attributes:
+      label: Description
+      description: "A clear description of what the bug is."
+    validations:
+      required: true
+
+  - type: textarea
+    id: steps
+    attributes:
+      label: Steps to Reproduce
+      description: "Step-by-step instructions to reproduce the behavior."
+      placeholder: |
+        1. Go to '...'
+        2. Click on '...'
+        3. See error
+    validations:
+      required: true
+
+  - type: textarea
+    id: expected
+    attributes:
+      label: Expected Behavior
+      description: "What did you expect to happen?"
+    validations:
+      required: true
+
+  - type: textarea
+    id: actual
+    attributes:
+      label: Actual Behavior
+      description: "What actually happened?"
+    validations:
+      required: true
+
+  - type: dropdown
+    id: test-impact
+    attributes:
+      label: Test Impact
+      description: "What automated test coverage should exist for this bug?"
+      options:
+        - Needs a new unit test
+        - Needs a new integration test
+        - Needs a new e2e test
+        - Existing automated test already fails
+        - Unsure
+    validations:
+      required: true
+
+  - type: textarea
+    id: logs
+    attributes:
+      label: Error Logs / Output
+      description: "Paste any relevant error messages, logs, or terminal output. This will be automatically formatted as code."
+      render: shell
+    validations:
+      required: false
+
+  - type: textarea
+    id: screenshots
+    attributes:
+      label: Screenshots
+      description: "If applicable, add screenshots to help explain the problem. Please also include the text of any error messages above — screenshots alone are not searchable."
+    validations:
+      required: false
+
+  - type: textarea
+    id: additional
+    attributes:
+      label: Additional Context
+      description: "Any other context about the problem (e.g. proxy config, number of accounts, network setup)."
+    validations:
+      required: false
+
+  - type: textarea
+    id: validation-plan
+    attributes:
+      label: Validation Plan
+      description: "Which commands or tests should prove this bug is fixed?"
+      placeholder: |
+        Example:
+        - node --import tsx --test tests/unit/my-file.test.ts
+        - npm run test:coverage
+    validations:
+      required: false

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,5 @@
+blank_issues_enabled: false
+contact_links:
+  - name: Question / Help
+    url: https://github.com/diegosouzapw/OmniRoute/discussions
+    about: For questions or help with setup, please use GitHub Discussions instead of opening an issue.

.github/ISSUE_TEMPLATE/feature_request.yml

@@ -0,0 +1,96 @@
+name: Feature Request
+description: Suggest a new feature or improvement for OmniRoute
+title: "[Feature] "
+labels: ["enhancement"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Thanks for suggesting a feature! Please describe the problem you're trying to solve and how you'd like it to work.
+
+  - type: textarea
+    id: problem
+    attributes:
+      label: Problem / Use Case
+      description: "What problem does this feature solve? Why do you need it?"
+      placeholder: "I'm trying to ... but currently ..."
+    validations:
+      required: true
+
+  - type: textarea
+    id: solution
+    attributes:
+      label: Proposed Solution
+      description: "How would you like this to work?"
+    validations:
+      required: true
+
+  - type: textarea
+    id: alternatives
+    attributes:
+      label: Alternatives Considered
+      description: "Have you considered any workarounds or alternative approaches?"
+    validations:
+      required: false
+
+  - type: textarea
+    id: acceptance
+    attributes:
+      label: Acceptance Criteria
+      description: "Describe the concrete behaviors or outcomes that should be validated before this is considered done."
+      placeholder: |
+        Example:
+        - API route returns 200 with valid payload
+        - Unit coverage added for the new branch
+        - Existing integrations remain green
+    validations:
+      required: true
+
+  - type: dropdown
+    id: area
+    attributes:
+      label: Area
+      description: "Which part of OmniRoute does this relate to?"
+      multiple: true
+      options:
+        - Dashboard / UI
+        - Proxy / Routing
+        - Provider Support
+        - CLI Tools Integration
+        - OAuth / Authentication
+        - Analytics / Usage Tracking
+        - Docker / Deployment
+        - Documentation
+        - Other
+    validations:
+      required: true
+
+  - type: input
+    id: provider
+    attributes:
+      label: Related Provider(s)
+      description: "If this relates to specific providers, list them."
+      placeholder: "e.g. Antigravity, OpenRouter, Ollama"
+    validations:
+      required: false
+
+  - type: textarea
+    id: additional
+    attributes:
+      label: Additional Context
+      description: "Any other context, mockups, or references."
+    validations:
+      required: false
+
+  - type: textarea
+    id: test-plan
+    attributes:
+      label: Expected Test Plan
+      description: "Which automated tests or coverage changes should accompany this work?"
+      placeholder: |
+        Example:
+        - Add unit tests for open-sse/services/combo.ts
+        - Extend integration coverage for /api/v1/models
+        - Keep npm run test:coverage at 60%+
+    validations:
+      required: false

.github/ISSUE_TEMPLATE/test_coverage_task.yml

@@ -0,0 +1,73 @@
+name: Test Coverage Task
+description: Create a focused coverage-improvement issue tied to concrete files and targets
+title: "[Coverage] "
+labels: ["test", "coverage"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Use this template for scoped coverage work. Keep it tied to specific files, measurable targets, and the gate that must stay green.
+
+  - type: input
+    id: baseline
+    attributes:
+      label: Current Coverage Baseline
+      description: "Paste the current overall or file-level coverage number that justifies this task."
+      placeholder: "e.g. Lines 79.00%, Branches 72.85%, open-sse/handlers/chatCore.ts = 67.22%"
+    validations:
+      required: true
+
+  - type: textarea
+    id: scope
+    attributes:
+      label: Target Files Or Modules
+      description: "List the concrete source files or directories that this task will cover."
+      placeholder: |
+        Example:
+        - open-sse/handlers/chatCore.ts
+        - open-sse/services/combo.ts
+        - tests/integration/chat-pipeline.test.ts
+    validations:
+      required: true
+
+  - type: textarea
+    id: scenarios
+    attributes:
+      label: Missing Scenarios
+      description: "Describe the specific behaviors, branches, or failure paths that are currently uncovered."
+      placeholder: |
+        Example:
+        - Upstream timeout path
+        - Empty tool_calls normalization
+        - Fallback route after first provider failure
+    validations:
+      required: true
+
+  - type: input
+    id: target
+    attributes:
+      label: Coverage Target
+      description: "Set the expected target for this task."
+      placeholder: "e.g. Raise open-sse/handlers to 80%+ lines and keep global gate >= 60%"
+    validations:
+      required: true
+
+  - type: textarea
+    id: validation
+    attributes:
+      label: Validation Commands
+      description: "List the commands that must pass before this issue can be closed."
+      placeholder: |
+        Example:
+        - node --import tsx --test tests/unit/my-suite.test.ts
+        - npm run test:coverage
+    validations:
+      required: true
+
+  - type: textarea
+    id: notes
+    attributes:
+      label: Notes
+      description: "Optional context, blockers, or dependencies."
+    validations:
+      required: false

.github/actions/npm-ci-retry/action.yml

@@ -0,0 +1,29 @@
+name: npm ci with retry
+description: Run npm ci with retries for transient registry/network failures.
+runs:
+  using: composite
+  steps:
+    - shell: bash
+      run: |
+        set -euo pipefail
+
+        max_attempts=3
+        delay_seconds=20
+
+        for attempt in $(seq 1 "$max_attempts"); do
+          if [ "$attempt" -gt 1 ]; then
+            echo "npm ci attempt $attempt/$max_attempts after transient failure"
+          fi
+
+          if npm ci; then
+            exit 0
+          fi
+
+          exit_code=$?
+          if [ "$attempt" -eq "$max_attempts" ]; then
+            exit "$exit_code"
+          fi
+
+          sleep "$delay_seconds"
+          delay_seconds=$((delay_seconds * 2))
+        done

.github/copilot-instructions.md

@@ -0,0 +1,15 @@
+# OmniRoute PR and Coverage Instructions
+
+- Treat `npm run test:coverage` as a required gate for PR work.
+- The repository minimum is `60%` for statements, lines, functions, and branches.
+- If a PR changes production code in `src/`, `open-sse/`, `electron/`, or `bin/`, it must include automated tests in the same PR.
+- When reviewing or updating a PR, if the report shows missing tests or coverage below `60%`, do not stop after reporting the problem. Add or update tests in the PR first, rerun the coverage gate, and only then ask for confirmation.
+- Prefer the smallest test layer that proves the behavior:
+  - unit tests first
+  - integration tests when multiple modules or DB state are involved
+  - e2e only when the behavior is truly UI or workflow-dependent
+- For bug issues, try to encode the reproduction as an automated test before or alongside the fix.
+- In the final PR report, include:
+  - the commands you ran
+  - the changed test files
+  - the final coverage result

.github/dependabot.yml

@@ -24,8 +24,38 @@ updates:
         update-types: ["version-update:semver-major"]
       - dependency-name: "eslint-config-next"
         update-types: ["version-update:semver-major"]
+      # jscpd v5 is a Rust rewrite (native binary, no Node.js programmatic API).
+      # scripts/check/check-duplication.mjs is deliberately pinned to jscpd@4 (it
+      # parses jscpd-report.json against a frozen baseline). A v5 major would break
+      # the duplication gate — migrate the gate intentionally, not via dependabot.
+      - dependency-name: "jscpd"
+        update-types: ["version-update:semver-major"]
+      # @huggingface/transformers is HARD-PINNED at 3.5.2 (exact, no caret) — FROZEN.
+      # It is load-bearing for the LLMLingua ONNX compression engine (open-sse/services/
+      # compression/engines/llmlingua/ — worker.ts pins @huggingface/transformers@3.5.2)
+      # and for local memory embeddings (src/lib/memory/embedding/transformersLocal.ts),
+      # and was VPS-validated at 3.5.2 (#4014). 4.x breaks both, and even 3.x minors must
+      # be re-validated on the VPS — so freeze ALL auto-bumps (no update-types = ignore
+      # every version). Migrate it intentionally, not via dependabot (#4050).
+      - dependency-name: "@huggingface/transformers"
 
   - package-ecosystem: "github-actions"
     directory: "/"
     schedule:
       interval: "weekly"
+
+  - package-ecosystem: "npm"
+    directory: "/electron"
+    schedule:
+      interval: "weekly"
+      day: "monday"
+    commit-message:
+      prefix: "deps"
+
+  - package-ecosystem: "docker"
+    directory: "/"
+    schedule:
+      interval: "weekly"
+      day: "monday"
+    commit-message:
+      prefix: "deps"

.github/pull_request_template.md

@@ -0,0 +1,30 @@
+## Summary
+
+- Describe the user-facing or operational change.
+
+## Related Issues
+
+- Closes #
+- Related to #
+
+## Validation
+
+- [ ] `npm run lint`
+- [ ] `npm run test:unit`
+- [ ] `npm run test:coverage`
+- [ ] Coverage is still `>= 60%` for statements, lines, functions, and branches
+- [ ] SonarQube PR analysis is green or any remaining issues are explicitly documented below
+
+## Tests Added Or Updated
+
+- List every changed or added automated test file.
+- If no production code changed, state that here.
+
+## Coverage Notes
+
+- If this PR changes `src/`, `open-sse/`, `electron/`, or `bin/`, explain which tests cover the change.
+- If coverage moved down in any touched file, explain why and what follow-up task will recover it.
+
+## Reviewer Notes
+
+- Call out any risky areas, migrations, feature flags, or manual validation that reviewers should know about.

.github/workflows/build-fork.yml

@@ -0,0 +1,71 @@
+name: Publish Fork Image to GHCR
+
+on:
+  push:
+    branches: [main]
+    tags:
+      - "v*"
+  workflow_dispatch:
+
+# Least-privilege default: read-only at the top level; the build job that pushes to
+# GHCR grants packages: write itself (Scorecard TokenPermissions).
+permissions:
+  contents: read
+
+env:
+  IMAGE_NAME: ghcr.io/kang-heewon/omniroute
+
+jobs:
+  build:
+    name: Build and Push Fork Image
+    if: github.repository == 'kang-heewon/OmniRoute'
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      packages: write
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@v4
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v4
+
+      - name: Login to GitHub Container Registry
+        uses: docker/login-action@v4
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Extract Docker metadata
+        id: meta
+        uses: docker/metadata-action@v6
+        with:
+          images: ${{ env.IMAGE_NAME }}
+          tags: |
+            type=raw,value=latest,enable={{is_default_branch}}
+            type=sha,prefix=sha-
+            type=ref,event=tag
+          labels: |
+            org.opencontainers.image.title=omniroute
+            org.opencontainers.image.description=Unified AI proxy/router — fork image
+            org.opencontainers.image.url=https://github.com/kang-heewon/OmniRoute
+            org.opencontainers.image.source=https://github.com/kang-heewon/OmniRoute
+            org.opencontainers.image.licenses=MIT
+
+      - name: Build and push
+        uses: docker/build-push-action@v7
+        with:
+          context: .
+          target: runner-base
+          platforms: linux/amd64,linux/arm64
+          push: true
+          tags: ${{ steps.meta.outputs.tags }}
+          labels: ${{ steps.meta.outputs.labels }}
+          cache-from: type=gha
+          cache-to: type=gha,mode=max

.github/workflows/claude.yml

@@ -0,0 +1,54 @@
+name: Claude Code
+
+on:
+  issue_comment:
+    types: [created]
+  pull_request_review_comment:
+    types: [created]
+  issues:
+    types: [opened, assigned]
+  pull_request_review:
+    types: [submitted]
+
+# Least-privilege default: no token permissions at the top level; the `claude` job
+# grants exactly what it needs below (Scorecard TokenPermissions).
+permissions: {}
+
+jobs:
+  claude:
+    if: |
+      (github.event_name == 'issue_comment' && contains(github.event.comment.body, '@claude')) ||
+      (github.event_name == 'pull_request_review_comment' && contains(github.event.comment.body, '@claude')) ||
+      (github.event_name == 'pull_request_review' && contains(github.event.review.body, '@claude')) ||
+      (github.event_name == 'issues' && (contains(github.event.issue.body, '@claude') || contains(github.event.issue.title, '@claude')))
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      pull-requests: read
+      issues: read
+      id-token: write
+      actions: read # Required for Claude to read CI results on PRs
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+          fetch-depth: 1
+
+      - name: Run Claude Code
+        id: claude
+        uses: anthropics/claude-code-action@v1
+        with:
+          claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
+
+          # This is an optional setting that allows Claude to read CI results on PRs
+          additional_permissions: |
+            actions: read
+
+          # Optional: Give a custom prompt to Claude. If this is not specified, Claude will perform the instructions specified in the comment that tagged it.
+          # prompt: 'Update the pull request description to include a summary of changes.'
+
+          # Optional: Add claude_args to customize behavior and configuration
+          # See https://github.com/anthropics/claude-code-action/blob/main/docs/usage.md
+          # or https://code.claude.com/docs/en/cli-reference for available options
+          # claude_args: '--allowed-tools Bash(gh pr *)'

.github/workflows/codeql.yml

@@ -0,0 +1,31 @@
+name: CodeQL
+# OWNER ACTION REQUIRED before enabling auto-triggers: advanced CodeQL conflicts with
+# GitHub "default setup" — the analyze step fails with "CodeQL analyses from advanced
+# configurations cannot be processed when the default setup is enabled". Switch repo
+# Settings → Code security → CodeQL from Default to Advanced, THEN restore the
+# push/pull_request/schedule triggers below. Until then this only runs on manual dispatch
+# so it never produces a red check on PRs. (The codeqlAlerts ratchet keeps working via the
+# default setup's alerts in the meantime.)
+on:
+  workflow_dispatch:
+permissions:
+  contents: read
+jobs:
+  analyze:
+    name: Analyze (javascript-typescript)
+    runs-on: ubuntu-latest
+    permissions:
+      security-events: write
+      actions: read
+      contents: read
+    steps:
+      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        with:
+          persist-credentials: false
+      - uses: github/codeql-action/init@8aad20d150bbac5944a9f9d289da16a4b0d87c1e # v4.36.2
+        with:
+          languages: javascript-typescript
+          queries: security-extended
+      - uses: github/codeql-action/analyze@8aad20d150bbac5944a9f9d289da16a4b0d87c1e # v4.36.2
+        with:
+          category: "/language:javascript-typescript"

.github/workflows/codex-review.yml

@@ -1,22 +0,0 @@
-name: Codex PR Review
-
-on:
-  pull_request:
-    types: [opened, synchronize]
-
-jobs:
-  request-codex-review:
-    runs-on: ubuntu-latest
-    permissions:
-      pull-requests: write
-    steps:
-      - name: Request Codex Review
-        uses: actions/github-script@v8
-        with:
-          script: |
-            await github.rest.issues.createComment({
-              owner: context.repo.owner,
-              repo: context.repo.repo,
-              issue_number: context.payload.pull_request.number,
-              body: '@codex review'
-            });

.github/workflows/dast-smoke.yml

@@ -0,0 +1,62 @@
+name: DAST smoke (PR)
+on:
+  pull_request:
+    branches: ["main", "release/**"]
+permissions:
+  contents: read
+jobs:
+  dast-smoke:
+    runs-on: ubuntu-latest
+    # ADVISORY while this new gate matures (repo convention: advisory -> blocking).
+    # Flip to blocking (remove continue-on-error) once it's proven stable across a few PRs.
+    continue-on-error: true
+    timeout-minutes: 12
+    env:
+      JWT_SECRET: ci-build-secret-with-sufficient-length-for-validation
+      API_KEY_SECRET: ci-api-key-secret-with-sufficient-length-aaaa
+    steps:
+      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@48b55a011bda9f5d6aeb4c2d9c7362e8dae4041e # v6.4.0
+        with:
+          node-version: "24"
+          cache: npm
+      - run: npm ci
+      - name: Build CLI bundle
+        run: npm run build:cli
+      - name: Start OmniRoute
+        env:
+          PORT: "20128"
+          INJECTION_GUARD_MODE: block
+        run: |
+          node dist/server.js > server.log 2>&1 &
+          echo $! > server.pid
+          for _ in $(seq 1 30); do
+            if curl -sf http://localhost:20128/api/monitoring/health >/dev/null; then echo up; break; fi
+            sleep 2
+          done
+      - uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
+        with:
+          python-version: "3.12"
+      - run: pip install schemathesis
+      - name: Schemathesis smoke (high-risk endpoints, blocking)
+        run: |
+          schemathesis run docs/openapi.yaml --url http://localhost:20128 \
+            --include-path-regex '^/v1/(chat/completions|models)$|^/api/(auth|keys)' \
+            --max-examples 8 --workers 4 --checks all --max-response-time 30 \
+            --request-timeout 20 --suppress-health-check all --no-color
+      - name: promptfoo injection-guard (blocking)
+        env:
+          OMNIROUTE_URL: http://localhost:20128
+          OMNIROUTE_API_KEY: not-needed-blocked-before-upstream
+        run: npx --yes promptfoo@latest eval -c promptfooconfig.yaml --no-cache
+      - name: Stop server
+        if: always()
+        run: kill "$(cat server.pid)" || true
+      - uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        if: always()
+        with:
+          name: dast-smoke-logs
+          path: server.log
+          retention-days: 7

.github/workflows/deploy-vps.yml

@@ -0,0 +1,98 @@
+name: Deploy to VPS
+
+on:
+  workflow_run:
+    workflows: ["Publish to Docker Hub"]
+    types: [completed]
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  deploy:
+    if: >-
+      (github.event_name == 'workflow_dispatch' || github.event.workflow_run.conclusion == 'success')
+      && vars.DEPLOY_ENABLED == 'true'
+    name: Deploy OmniRoute to VPS
+    runs-on: ubuntu-latest
+    steps:
+      - name: Check VPS SSH reachability from runner
+        id: reach
+        env:
+          # Pass the host via env (never interpolate a secret straight into the
+          # script body) so /dev/tcp gets a shell variable, not inlined text.
+          VPS_HOST: ${{ secrets.VPS_HOST }}
+        run: |
+          set -uo pipefail
+          # A GitHub-hosted runner can only deploy when it can actually open a TCP
+          # connection to the VPS SSH port. The Local VPS lives on a private LAN and
+          # the Akamai host firewalls :22 to known IPs, so the runner is routinely
+          # unable to reach it (`dial tcp ***:22: i/o timeout`). Treat "unreachable
+          # from the runner" as a SKIP — the real deploys are run manually from an
+          # allowed network via the deploy-vps-local / deploy-vps-akamai skills — so
+          # an unreachable host no longer red-fails every release/push pipeline.
+          # When the host IS reachable, the deploy step below still runs in full and
+          # its health gate surfaces any genuine deploy failure.
+          if timeout 15 bash -c 'exec 3<>"/dev/tcp/${VPS_HOST}/22"' 2>/dev/null; then
+            echo "reachable=true" >> "$GITHUB_OUTPUT"
+            echo "✅ VPS_HOST:22 reachable from the runner — proceeding with deploy."
+          else
+            echo "reachable=false" >> "$GITHUB_OUTPUT"
+            echo "::warning title=Auto-deploy skipped::VPS_HOST:22 is not reachable from this GitHub runner (private LAN / firewalled). Deploy manually with the deploy-vps-local or deploy-vps-akamai skill."
+          fi
+
+      - name: Deploy via SSH
+        if: steps.reach.outputs.reachable == 'true'
+        uses: appleboy/ssh-action@v1
+        with:
+          host: ${{ secrets.VPS_HOST }}
+          username: ${{ secrets.VPS_USER }}
+          key: ${{ secrets.VPS_SSH_KEY }}
+          port: 22
+          timeout: 60s
+          command_timeout: 15m
+          script: |
+            set -euo pipefail
+
+            echo "=== Updating OmniRoute ==="
+            npm install -g omniroute@latest
+            INSTALLED_VERSION=$(omniroute --version 2>/dev/null | tr -d '[:space:]' || echo "unknown")
+            echo "Installed CLI version: $INSTALLED_VERSION"
+
+            # Recreate the PM2 process instead of `pm2 restart`. A bare restart
+            # re-runs whatever script path was saved earlier; after the build-output
+            # reorg (app/ -> dist/, .next -> .build/next) a process pinned to the old
+            # app/server-ws.mjs path can no longer start, and the node process dies
+            # while PM2 still reports "online" — so the box never binds :20128.
+            # Always launch via the `omniroute` bin so .env is loaded and the dist/
+            # layout is resolved correctly.
+            echo "=== (Re)creating PM2 process via bin ==="
+            pm2 delete omniroute 2>/dev/null || true
+            pm2 start omniroute --name omniroute -- --port 20128
+            pm2 save
+
+            # Health gate: fail the deploy unless the box actually reports healthy.
+            # Poll /api/monitoring/health for "status":"healthy" (a deeper signal than
+            # a static page 200 — it confirms the app booted, not just that a port is
+            # bound). Boot can take a while after a native-module/build-layout change,
+            # so poll up to ~3min before giving up.
+            echo "=== Health Check (gates the deploy) ==="
+            ok=0
+            for i in $(seq 1 36); do
+              BODY=$(curl -sf -m 5 http://localhost:20128/api/monitoring/health 2>/dev/null || true)
+              if printf '%s' "$BODY" | grep -q '"status":"healthy"'; then
+                ok=1
+                echo "✅ /api/monitoring/health -> healthy (attempt $i) — version $INSTALLED_VERSION"
+                break
+              fi
+              echo "… not healthy yet (attempt $i/36), retrying in 5s"
+              sleep 5
+            done
+            if [ "$ok" != "1" ]; then
+              echo "❌ Health check failed — /api/monitoring/health never reported healthy after ~3min"
+              echo "--- recent PM2 logs ---"
+              pm2 logs omniroute --lines 40 --nostream || true
+              exit 1
+            fi
+            echo "=== Deploy complete ==="

.github/workflows/docker-publish.yml

@@ -1,51 +1,391 @@
 name: Publish to Docker Hub
 
 on:
+  push:
+    branches:
+      - main
+    tags:
+      - "v*"
+    paths-ignore:
+      - ".github/workflows/**"
+  # Use 'released' instead of 'published' so editing/re-publishing old releases
+  # does NOT re-trigger this workflow. 'released' fires only on the initial
+  # release publication (and pre-release → release transition).
   release:
-    types: [published]
+    types: [released]
+  workflow_dispatch:
+    inputs:
+      version:
+        description: "Version tag to build (e.g. 3.8.4)"
+        required: true
+        type: string
+      promote_latest:
+        description: "Also tag :latest (only if this is the highest semver)"
+        required: false
+        type: boolean
+        default: false
 
+# Least-privilege default: read-only at the top level; the build and merge jobs that
+# push to GHCR grant packages: write themselves (Scorecard TokenPermissions).
 permissions:
   contents: read
 
 jobs:
-  docker:
-    name: Build & Push Docker Image
+  prepare:
+    name: Resolve Docker release metadata
     runs-on: ubuntu-latest
+    outputs:
+      version: ${{ steps.version.outputs.version }}
+      promote_latest: ${{ steps.version.outputs.promote_latest }}
+      skip: ${{ steps.version.outputs.skip }}
+    env:
+      IMAGE_NAME: diegosouzapw/omniroute
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+          ref: ${{ github.event_name == 'workflow_dispatch' && format('refs/tags/v{0}', inputs.version) || '' }}
+          # Need full tag history for semver comparison when deciding :latest.
+          fetch-depth: 0
 
-      - name: Extract version from release tag
+      - name: Resolve version, latest-promotion, and skip flag
         id: version
+        env:
+          EVENT_NAME: ${{ github.event_name }}
+          REF_NAME: ${{ github.ref_name }}
+          REF_TYPE: ${{ github.ref_type }}
+          INPUT_VERSION: ${{ inputs.version }}
+          PROMOTE_INPUT: ${{ inputs.promote_latest }}
         run: |
-          VERSION="${GITHUB_REF_NAME}"
-          VERSION="${VERSION#v}"
+          set -euo pipefail
+
+          # 1) Resolve version string from the trigger (all inputs come via env).
+          case "$EVENT_NAME" in
+            workflow_dispatch)
+              VERSION="${INPUT_VERSION#v}"
+              ;;
+            push)
+              if [ "$REF_TYPE" = "tag" ]; then
+                VERSION="${REF_NAME#v}"
+              else
+                # Push to main → build & tag as `main` only. Never touch :latest.
+                VERSION="main"
+              fi
+              ;;
+            release)
+              VERSION="${REF_NAME#v}"
+              ;;
+            *)
+              VERSION="${REF_NAME#v}"
+              ;;
+          esac
+          # Sanity-check: only allow [A-Za-z0-9._-] in VERSION (defense in depth).
+          if ! printf '%s' "$VERSION" | grep -qE '^[A-Za-z0-9._-]+$'; then
+            echo "Refusing to use unsafe VERSION value: $VERSION" >&2
+            exit 1
+          fi
           echo "version=$VERSION" >> "$GITHUB_OUTPUT"
-          echo "Publishing Docker image version: $VERSION"
+
+          # 2) Decide whether to promote :latest.
+          PROMOTE="false"
+          if [ "$VERSION" = "main" ]; then
+            PROMOTE="false"
+          elif printf '%s' "$VERSION" | grep -qE -- '-(rc|alpha|beta|pre|next)'; then
+            echo "Pre-release identifier detected — skipping :latest."
+            PROMOTE="false"
+          elif [ "$EVENT_NAME" = "workflow_dispatch" ]; then
+            PROMOTE="${PROMOTE_INPUT:-false}"
+          else
+            git fetch --tags --quiet || true
+            HIGHEST=$(git tag -l 'v[0-9]*' | sed 's/^v//' | grep -vE -- '-(rc|alpha|beta|pre|next)' | sort -V | tail -1 || echo "")
+            if [ -n "$HIGHEST" ] && [ "$VERSION" = "$HIGHEST" ]; then
+              PROMOTE="true"
+            else
+              echo "Version $VERSION is not the highest semver tag (highest=${HIGHEST:-<none>}). Not promoting :latest."
+            fi
+          fi
+          echo "promote_latest=$PROMOTE" >> "$GITHUB_OUTPUT"
+
+          # 3) Skip if this exact version is already published in Docker Hub.
+          #    `main` is always rebuilt (mutable floating tag).
+          SKIP="false"
+          if [ "$VERSION" != "main" ]; then
+            if docker manifest inspect "diegosouzapw/omniroute:${VERSION}" >/dev/null 2>&1; then
+              echo "Image diegosouzapw/omniroute:${VERSION} already exists on Docker Hub — skipping rebuild."
+              SKIP="true"
+            fi
+          fi
+          echo "skip=$SKIP" >> "$GITHUB_OUTPUT"
+
+          echo "Publishing diegosouzapw/omniroute:$VERSION (promote_latest=$PROMOTE, skip=$SKIP)"
+
+  build:
+    name: Build Docker (${{ matrix.platform }})
+    needs: prepare
+    if: needs.prepare.outputs.skip != 'true'
+    runs-on: ${{ matrix.runner }}
+    permissions:
+      contents: read
+      packages: write
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - platform: linux/amd64
+            runner: ubuntu-24.04
+            arch: amd64
+          - platform: linux/arm64
+            runner: ubuntu-24.04-arm
+            arch: arm64
+    env:
+      IMAGE_NAME: diegosouzapw/omniroute
+      GHCR_IMAGE_NAME: ghcr.io/diegosouzapw/omniroute
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+          ref: ${{ github.event_name == 'workflow_dispatch' && format('refs/tags/v{0}', inputs.version) || '' }}
+          fetch-depth: 0
 
       - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
+        uses: docker/setup-buildx-action@v4
 
       - name: Login to Docker Hub
-        uses: docker/login-action@v3
+        uses: docker/login-action@v4
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
 
-      - name: Build and push
-        uses: docker/build-push-action@v6
+      - name: Login to GitHub Container Registry
+        uses: docker/login-action@v4
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Build and push platform image by digest
+        id: build
+        uses: docker/build-push-action@v7
         with:
           context: .
           target: runner-base
-          push: true
+          platforms: ${{ matrix.platform }}
+          outputs: type=image,push-by-digest=true,name-canonical=true,push=true
           tags: |
-            diegosouzapw/omniroute:${{ steps.version.outputs.version }}
-            diegosouzapw/omniroute:latest
-          cache-from: type=gha
-          cache-to: type=gha,mode=max
-          platforms: linux/amd64
+            ${{ env.IMAGE_NAME }}
+            ${{ env.GHCR_IMAGE_NAME }}
+          cache-from: type=gha,scope=docker-${{ matrix.arch }}
+          cache-to: type=gha,scope=docker-${{ matrix.arch }},mode=max
+          no-cache: false
+        env:
+          DOCKER_BUILDKIT_INLINE_CACHE: 1
+
+      - name: Build and push WEB platform image by digest
+        id: build-web
+        uses: docker/build-push-action@v7
+        with:
+          context: .
+          target: runner-web
+          platforms: ${{ matrix.platform }}
+          outputs: type=image,push-by-digest=true,name-canonical=true,push=true
+          tags: |
+            ${{ env.IMAGE_NAME }}
+            ${{ env.GHCR_IMAGE_NAME }}
+          cache-from: type=gha,scope=docker-web-${{ matrix.arch }}
+          cache-to: type=gha,scope=docker-web-${{ matrix.arch }},mode=max
+          no-cache: false
+        env:
+          DOCKER_BUILDKIT_INLINE_CACHE: 1
+
+      - name: Export digests
+        env:
+          DIGEST_BASE: ${{ steps.build.outputs.digest }}
+          DIGEST_WEB: ${{ steps.build-web.outputs.digest }}
+        run: |
+          set -euo pipefail
+          mkdir -p /tmp/digests/base /tmp/digests/web
+          touch "/tmp/digests/base/${DIGEST_BASE#sha256:}"
+          touch "/tmp/digests/web/${DIGEST_WEB#sha256:}"
+
+      - name: Upload base digests
+        uses: actions/upload-artifact@v7
+        with:
+          name: digests-base-${{ matrix.arch }}
+          path: /tmp/digests/base/*
+          if-no-files-found: error
+          retention-days: 1
+
+      - name: Upload web digests
+        uses: actions/upload-artifact@v7
+        with:
+          name: digests-web-${{ matrix.arch }}
+          path: /tmp/digests/web/*
+          if-no-files-found: error
+          retention-days: 1
+
+  merge:
+    name: Publish multi-arch manifests
+    needs:
+      - prepare
+      - build
+    if: needs.prepare.outputs.skip != 'true'
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      packages: write
+      security-events: write
+    env:
+      IMAGE_NAME: diegosouzapw/omniroute
+      GHCR_IMAGE_NAME: ghcr.io/diegosouzapw/omniroute
+      VERSION: ${{ needs.prepare.outputs.version }}
+      PROMOTE_LATEST: ${{ needs.prepare.outputs.promote_latest }}
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+          ref: ${{ github.event_name == 'workflow_dispatch' && format('refs/tags/v{0}', inputs.version) || '' }}
+          fetch-depth: 0
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v4
+
+      - name: Login to Docker Hub
+        uses: docker/login-action@v4
+        with:
+          username: ${{ secrets.DOCKERHUB_USERNAME }}
+          password: ${{ secrets.DOCKERHUB_TOKEN }}
+
+      - name: Login to GitHub Container Registry
+        uses: docker/login-action@v4
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Download base digests
+        uses: actions/download-artifact@v8
+        with:
+          pattern: digests-base-*
+          path: /tmp/digests/base
+          merge-multiple: true
+
+      - name: Download web digests
+        uses: actions/download-artifact@v8
+        with:
+          pattern: digests-web-*
+          path: /tmp/digests/web
+          merge-multiple: true
+
+      - name: Create Docker Hub manifest
+        run: |
+          set -euo pipefail
+
+          create_manifest() {
+            local image="$1" suffix="$2" dir="$3"
+            local tags=(-t "${image}:${VERSION}${suffix}")
+            if [ "$PROMOTE_LATEST" = "true" ]; then
+              tags+=(-t "${image}:latest${suffix}")
+            fi
+            local refs=()
+            while IFS= read -r digest_file; do
+              refs+=("${image}@sha256:$(basename "$digest_file")")
+            done < <(find "$dir" -type f | sort)
+            if [ "${#refs[@]}" -eq 0 ]; then
+              echo "No image digests in $dir" >&2
+              exit 1
+            fi
+            docker buildx imagetools create "${tags[@]}" "${refs[@]}"
+          }
+
+          create_manifest "${IMAGE_NAME}" "" /tmp/digests/base
+          create_manifest "${IMAGE_NAME}" "-web" /tmp/digests/web
+
+      - name: Create GHCR manifest
+        run: |
+          set -euo pipefail
+
+          create_manifest() {
+            local image="$1" suffix="$2" dir="$3"
+            local tags=(-t "${image}:${VERSION}${suffix}")
+            if [ "$PROMOTE_LATEST" = "true" ]; then
+              tags+=(-t "${image}:latest${suffix}")
+            fi
+            local refs=()
+            while IFS= read -r digest_file; do
+              refs+=("${image}@sha256:$(basename "$digest_file")")
+            done < <(find "$dir" -type f | sort)
+            if [ "${#refs[@]}" -eq 0 ]; then
+              echo "No image digests in $dir" >&2
+              exit 1
+            fi
+            docker buildx imagetools create "${tags[@]}" "${refs[@]}"
+          }
+
+          create_manifest "${GHCR_IMAGE_NAME}" "" /tmp/digests/base
+          create_manifest "${GHCR_IMAGE_NAME}" "-web" /tmp/digests/web
+
+      - name: Inspect image
+        if: needs.prepare.outputs.version != 'main'
+        run: |
+          docker buildx imagetools inspect "${IMAGE_NAME}:${VERSION}"
+
+      - name: Generate CycloneDX SBOM (image, advisory)
+        if: needs.prepare.outputs.version != 'main'
+        continue-on-error: true
+        uses: anchore/sbom-action@v0
+        with:
+          image: ${{ env.GHCR_IMAGE_NAME }}:${{ env.VERSION }}
+          format: cyclonedx-json
+          output-file: sbom-image.cdx.json
+          artifact-name: sbom-image.cdx.json
+
+      # Visibility scan: reports HIGH + CRITICAL into the SARIF (Security tab) but
+      # never blocks (exit-code 0). The blocking gate below narrows to CRITICAL.
+      - name: Trivy image scan (SARIF, advisory)
+        if: needs.prepare.outputs.version != 'main'
+        continue-on-error: true
+        uses: aquasecurity/trivy-action@ed142fd0673e97e23eac54620cfb913e5ce36c25 # v0.36.0
+        with:
+          image-ref: ${{ env.GHCR_IMAGE_NAME }}:${{ env.VERSION }}
+          format: sarif
+          output: trivy-results.sarif
+          severity: HIGH,CRITICAL
+          exit-code: "0"
+
+      # BLOCKING gate (v3.8.27 cycle-end): fail the release on a CRITICAL CVE in the
+      # published image. Narrowed to severity CRITICAL (HIGH stays visible in the
+      # SARIF step above, not blocking). ignore-unfixed:true so an unfixable base-image
+      # CVE with no upstream patch does not red the release (reduces false-blocks);
+      # a fixable CRITICAL still blocks. Per docs/security/SUPPLY_CHAIN.md. NB: Trivy
+      # scans against a CVE DB that grows continuously — a newly-disclosed CRITICAL on
+      # an unchanged base image can red this gate; the fix is to rebuild on a patched
+      # base, bump the dep, or add a justified .trivyignore entry (see the CVE-variance
+      # note in docs/security/SUPPLY_CHAIN.md).
+      - name: Trivy CRITICAL gate (blocking)
+        if: needs.prepare.outputs.version != 'main'
+        uses: aquasecurity/trivy-action@ed142fd0673e97e23eac54620cfb913e5ce36c25 # v0.36.0
+        with:
+          image-ref: ${{ env.GHCR_IMAGE_NAME }}:${{ env.VERSION }}
+          format: table
+          severity: CRITICAL
+          ignore-unfixed: true
+          exit-code: "1"
+
+      - name: Upload Trivy SARIF to Security tab
+        if: needs.prepare.outputs.version != 'main'
+        continue-on-error: true
+        uses: github/codeql-action/upload-sarif@v4
+        with:
+          sarif_file: trivy-results.sarif
+          category: trivy-image
 
       - name: Update Docker Hub description
+        # Only refresh README/description when we actually promote :latest
+        # (avoids overwriting from main pushes or back-fill builds).
+        if: needs.prepare.outputs.promote_latest == 'true'
         uses: peter-evans/dockerhub-description@v5
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}

.github/workflows/electron-release.yml

@@ -0,0 +1,287 @@
+name: Build Electron Desktop App
+
+on:
+  push:
+    tags:
+      - "v*"
+  workflow_dispatch:
+    inputs:
+      version:
+        description: "Release version (e.g., v1.6.8)"
+        required: true
+        type: string
+
+# Least-privilege default: read-only at the top level; each job grants the writes it
+# needs (build/release upload assets, publish-npm forwards npm provenance / packages
+# to the reusable workflow) — Scorecard TokenPermissions.
+permissions:
+  contents: read
+
+jobs:
+  validate:
+    name: Validate version
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+    outputs:
+      version: ${{ steps.validate.outputs.version }}
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+          fetch-depth: 0
+
+      - name: Validate version format
+        id: validate
+        env:
+          # Pass workflow context via env (never interpolate ${{ ... }} straight
+          # into the run: script body) so the shell receives variables, not
+          # inlined text — zizmor template-injection mitigation. INPUT_VERSION is
+          # the operator-supplied value and is regex-validated below before use.
+          EVENT_NAME: ${{ github.event_name }}
+          INPUT_VERSION: ${{ inputs.version }}
+        run: |
+          if [[ "$EVENT_NAME" == "push" ]]; then
+            VERSION="${GITHUB_REF#refs/tags/}"
+          else
+            VERSION="$INPUT_VERSION"
+          fi
+
+          if [[ ! "$VERSION" =~ ^v[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
+            echo "Error: Invalid version format. Expected: v1.6.8"
+            exit 1
+          fi
+          echo "version=$VERSION" >> "$GITHUB_OUTPUT"
+          echo "✓ Valid version: $VERSION"
+
+  build:
+    name: Build Electron (${{ matrix.platform }})
+    needs: validate
+    runs-on: ${{ matrix.runner }}
+    permissions:
+      contents: write # electron-builder may publish artifacts with GH_TOKEN
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - platform: windows
+            runner: windows-latest
+            target: win
+            ext: .exe
+          - platform: macos-intel
+            runner: macos-15-intel
+            target: mac-x64
+            ext: .dmg
+          - platform: macos-arm64
+            runner: macos-latest
+            target: mac-arm64
+            ext: -arm64.dmg
+          - platform: linux
+            runner: ubuntu-latest
+            target: linux
+            ext: .AppImage
+            deb_ext: .deb
+
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - name: Setup Node
+        uses: actions/setup-node@v6
+        with:
+          node-version: 24
+          cache: npm
+
+      - name: Cache node_modules
+        uses: actions/cache@v5.0.5
+        with:
+          path: node_modules
+          key: ${{ runner.os }}-node-${{ hashFiles('package-lock.json') }}
+          restore-keys: |
+            ${{ runner.os }}-node-
+
+      - name: Install dependencies
+        run: npm ci
+        env:
+          NPM_CONFIG_LEGACY_PEER_DEPS: true
+
+      - name: Sanitize Windows home directory
+        if: runner.os == 'Windows'
+        shell: bash
+        run: |
+          # The default USERPROFILE contains junction points (Application Data)
+          # that cause EPERM errors during Next.js standalone build glob scans.
+          # Create a clean temp profile directory to avoid this.
+          mkdir -p "$RUNNER_TEMP/home"
+          echo "USERPROFILE=$RUNNER_TEMP/home" >> "$GITHUB_ENV"
+
+      - name: Build Next.js standalone
+        env:
+          JWT_SECRET: ci-build-secret-with-sufficient-length-for-validation
+          NODE_OPTIONS: "--max_old_space_size=6144"
+        run: npm run build
+
+      - name: Sync version in electron/package.json
+        shell: bash
+        env:
+          # Pass the validated version via env (never interpolate ${{ ... }}
+          # straight into the run: script body) — zizmor template-injection
+          # mitigation. Already regex-validated (^v[0-9]+\.[0-9]+\.[0-9]+$) in
+          # the `validate` job, so it cannot carry shell metacharacters.
+          VERSION: ${{ needs.validate.outputs.version }}
+        run: |
+          VERSION_NO_V="${VERSION#v}"
+          node -e "
+            const fs = require('fs');
+            const pkg = JSON.parse(fs.readFileSync('electron/package.json'));
+            pkg.version = '$VERSION_NO_V';
+            fs.writeFileSync('electron/package.json', JSON.stringify(pkg, null, 2) + '\\n');
+          "
+          echo "✓ electron/package.json version set to $VERSION_NO_V"
+
+      - name: Install fpm (Linux .deb packaging tool)
+        if: matrix.platform == 'linux'
+        run: sudo gem install fpm --no-document
+
+      - name: Install Electron dependencies
+        working-directory: electron
+        run: npm install --no-audit --no-fund
+
+      - name: Build Electron for ${{ matrix.platform }}
+        working-directory: electron
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: npm run build:${{ matrix.target }}
+
+      - name: Smoke packaged Electron app
+        if: matrix.platform != 'linux'
+        # Best-effort smoke on Windows + macos-arm64:
+        #  - Windows: requestSingleInstanceLock() fails due to USERPROFILE
+        #    sanitization needed for the build step.
+        #  - macos-arm64: the headless GitHub arm64 runner crashes Electron's GPU
+        #    process (gpu_process_host exit_code=15 → network service crash →
+        #    "No rendezvous client, terminating process"), so the app can't bind
+        #    127.0.0.1:20128 in time. The identical bundle is smoke-gated on
+        #    macos-intel + linux, so packaging is still verified per-OS; we don't
+        #    let the arm64 runner's GPU flakiness block the desktop release.
+        continue-on-error: ${{ matrix.platform == 'windows' || matrix.platform == 'macos-arm64' }}
+        env:
+          ELECTRON_SMOKE_TIMEOUT_MS: 60000
+          ELECTRON_SMOKE_STREAM_LOGS: "1"
+        run: npm run electron:smoke:packaged
+
+      - name: Smoke packaged Electron app (Linux)
+        if: matrix.platform == 'linux'
+        env:
+          ELECTRON_SMOKE_TIMEOUT_MS: 60000
+          ELECTRON_SMOKE_STREAM_LOGS: "1"
+        run: xvfb-run -a npm run electron:smoke:packaged
+
+      - name: Collect installers
+        shell: bash
+        run: |
+          mkdir -p release-assets
+          cd electron/dist-electron
+          # Copy only installer files for this platform
+          for file in *${{ matrix.ext }}; do
+            [ -f "$file" ] && cp "$file" ../../release-assets/
+          done
+          # Linux: also copy .deb package
+          if [ "${{ matrix.platform }}" = "linux" ]; then
+            for file in *.deb; do
+              [ -f "$file" ] && cp "$file" ../../release-assets/
+            done
+          fi
+          # Windows: also copy portable standalone exe as OmniRoute.exe
+          if [ "${{ matrix.platform }}" = "windows" ]; then
+            for file in *.exe; do
+              # Skip the NSIS installer (contains "Setup")
+              case "$file" in *Setup*) continue ;; esac
+              [ -f "$file" ] && cp "$file" "../../release-assets/OmniRoute.exe" && break
+            done
+          fi
+
+      - name: Upload artifacts
+        uses: actions/upload-artifact@v7
+        with:
+          name: electron-${{ matrix.platform }}
+          path: release-assets/
+
+  release:
+    name: Create Release
+    needs: [validate, build]
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write # softprops/action-gh-release creates the GitHub Release
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+          fetch-depth: 0
+
+      - name: Download all artifacts
+        uses: actions/download-artifact@v8
+        with:
+          path: release-assets
+          merge-multiple: true
+
+      - name: Create source archives
+        env:
+          # Pass the validated version via env (never interpolate ${{ ... }}
+          # straight into the run: script body) — zizmor template-injection
+          # mitigation. Already regex-validated (^v[0-9]+\.[0-9]+\.[0-9]+$) in
+          # the `validate` job, so it cannot carry shell metacharacters.
+          VERSION: ${{ needs.validate.outputs.version }}
+        run: |
+          # Create source code archives (excluding dev dependencies and build artifacts)
+          export TARBALL="OmniRoute-${VERSION}.source.tar.gz"
+          export ZIPBALL="OmniRoute-${VERSION}.source.zip"
+
+          # Use git archive for clean source export
+          git archive --format=tar.gz --prefix="OmniRoute-${VERSION}/" HEAD -o "release-assets/$TARBALL"
+          git archive --format=zip --prefix="OmniRoute-${VERSION}/" HEAD -o "release-assets/$ZIPBALL"
+
+          echo "✓ Created source archives:"
+          ls -lh "release-assets/$TARBALL" "release-assets/$ZIPBALL"
+
+      - name: List release files
+        run: ls -la release-assets/
+
+      - name: Create Release
+        uses: softprops/action-gh-release@v3
+        with:
+          tag_name: ${{ needs.validate.outputs.version }}
+          draft: false
+          prerelease: false
+          generate_release_notes: true
+          fail_on_unmatched_files: false
+          files: |
+            release-assets/*.dmg
+            release-assets/*.exe
+            release-assets/*.AppImage
+            release-assets/*.deb
+            release-assets/*.blockmap
+            release-assets/*.source.tar.gz
+            release-assets/*.source.zip
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+  publish-npm:
+    name: Publish to npm
+    needs: [validate, release]
+    permissions:
+      # Must be `write`, not `read`: this job calls the reusable npm-publish.yml whose
+      # `publish` job needs `contents: write` (gh release upload — attach the SBOM, #3874).
+      # A reusable workflow's job cannot request more permission than the caller grants,
+      # so a `read` here makes GitHub reject the run at startup (startup_failure).
+      contents: write
+      id-token: write # npm provenance (forwarded to the reusable workflow)
+      packages: write # publish to npm.pkg.github.com
+    uses: ./.github/workflows/npm-publish.yml
+    with:
+      version: ${{ needs.validate.outputs.version }}
+      tag: latest
+    secrets:
+      NPM_TOKEN: ${{ secrets.NPM_TOKEN }}

.github/workflows/lock-released-branch.yml

@@ -0,0 +1,125 @@
+name: Lock released branch
+
+# Two responsibilities (defense in depth — Hard Rule #18 enforcement):
+#
+# 1. `on: release: published` — when a GitHub Release publishes tag v3.X.Y,
+#    apply branch protection (lock_branch + enforce_admins) to release/v3.X.Y
+#    so no further commits can land on a shipped version. To reopen later:
+#      gh api -X DELETE repos/<owner>/<repo>/branches/release/<tag>/protection
+#
+# 2. `on: push: branches: ['release/v*']` — verify that no push lands on a
+#    release/* branch whose matching tag already exists. This is the preventive
+#    guard: if the lock didn't apply (workflow bug, missing PAT, race), this
+#    job FAILS the push run so the operator gets paged immediately.
+#
+# `permissions:` cannot grant the `Administration` scope to GITHUB_TOKEN — that
+# scope only exists on PATs. Set BRANCH_LOCK_TOKEN as a repo secret pointing to
+# a PAT/fine-grained token with `Administration: read & write`. Without it, the
+# lock step will fail loudly (which is what we want — silent failure caused the
+# v3.8.3 incident on 2026-05-26 where 6 commits landed post-release).
+
+on:
+  release:
+    types: [published]
+  push:
+    branches:
+      - "release/v*"
+  workflow_dispatch:
+    inputs:
+      tag:
+        description: "Tag of the released version (e.g. v3.8.2)"
+        required: true
+        type: string
+
+permissions:
+  contents: read
+
+jobs:
+  # ─────────────────────────────────────────────────────────────────────────
+  # Job 1 — Lock the release branch when a Release is published.
+  # ─────────────────────────────────────────────────────────────────────────
+  lock-branch:
+    if: github.event_name == 'release' || github.event_name == 'workflow_dispatch'
+    runs-on: ubuntu-latest
+    steps:
+      - name: Lock release/<tag> branch
+        env:
+          # Administration scope is required to PUT branch protection. Default
+          # GITHUB_TOKEN cannot do this — operator must provision BRANCH_LOCK_TOKEN.
+          GH_TOKEN: ${{ secrets.BRANCH_LOCK_TOKEN }}
+          TAG: ${{ github.event.release.tag_name || inputs.tag }}
+          REPO: ${{ github.repository }}
+        run: |
+          set -euo pipefail
+
+          if [ -z "${GH_TOKEN}" ]; then
+            echo "::error::BRANCH_LOCK_TOKEN secret is not set. Create a PAT with Administration:write and add it as repo secret."
+            exit 1
+          fi
+          if [ -z "${TAG}" ]; then
+            echo "::error::No tag provided; cannot determine release branch."
+            exit 1
+          fi
+
+          BRANCH="release/${TAG}"
+          echo "Target branch: ${BRANCH} (repo: ${REPO})"
+
+          if ! gh api "repos/${REPO}/branches/${BRANCH}" >/dev/null 2>&1; then
+            echo "::warning::Branch ${BRANCH} not found — nothing to lock."
+            exit 0
+          fi
+
+          echo "Applying lock_branch protection to ${BRANCH}..."
+          gh api -X PUT "repos/${REPO}/branches/${BRANCH}/protection" --input - <<'JSON'
+          {
+            "required_status_checks": null,
+            "enforce_admins": true,
+            "required_pull_request_reviews": null,
+            "restrictions": null,
+            "lock_branch": true,
+            "allow_force_pushes": false,
+            "allow_deletions": false
+          }
+          JSON
+
+          LOCKED=$(gh api "repos/${REPO}/branches/${BRANCH}/protection" \
+            --jq '.lock_branch.enabled')
+          if [ "${LOCKED}" != "true" ]; then
+            echo "::error::Failed to confirm lock on ${BRANCH} (lock_branch=${LOCKED})."
+            exit 1
+          fi
+          echo "✅ ${BRANCH} is now locked (read-only)."
+
+  # ─────────────────────────────────────────────────────────────────────────
+  # Job 2 — Preventive guard: fail if a push lands on release/vX.Y.Z whose
+  # tag already exists. This catches the case where the lock didn't apply
+  # (PAT missing, race window, workflow bug) and pages the operator.
+  # ─────────────────────────────────────────────────────────────────────────
+  guard-no-push-after-release:
+    if: github.event_name == 'push'
+    runs-on: ubuntu-latest
+    steps:
+      - name: Reject push if matching release tag exists
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          REPO: ${{ github.repository }}
+          REF: ${{ github.ref_name }}
+        run: |
+          set -euo pipefail
+
+          # Extract version from ref: release/v3.8.3 -> v3.8.3
+          if [[ ! "${REF}" =~ ^release/(v[0-9]+\.[0-9]+\.[0-9]+)$ ]]; then
+            echo "Ref ${REF} does not match release/vX.Y.Z — nothing to guard."
+            exit 0
+          fi
+          TAG="${BASH_REMATCH[1]}"
+          echo "Checking if tag ${TAG} already exists on ${REPO}..."
+
+          if gh api "repos/${REPO}/git/refs/tags/${TAG}" >/dev/null 2>&1; then
+            echo "::error::Hard Rule #18 violation — push to ${REF} but tag ${TAG} is already released."
+            echo "::error::Hotfixes for a released version must go on a NEW branch: release/v$(echo "${TAG#v}" | awk -F. '{$3=$3+1; print $1"."$2"."$3}' OFS=.)"
+            echo "::error::To undo this push: revert the offending commits, or contact an admin to lock the branch if it wasn't already."
+            exit 1
+          fi
+
+          echo "✅ No release tag for ${TAG} yet — push is OK."

.github/workflows/mutation-redundancy.yml

@@ -0,0 +1,63 @@
+name: Mutation Redundancy (disableBail, on-demand)
+# One-off measurement to UNBLOCK R1 (test-redundancy prune). The nightly mutation run
+# (nightly-mutation.yml) bails on the first kill, so `killedBy` lists only the FIRST
+# killer — 🟠 redundant is understated and 🟢 unique overstated (see the caveat in
+# scripts/quality/mutation-radiography.mjs). This workflow re-runs the SAME combo +
+# chatCore leaf batches with stryker.disablebail.json (disableBail:true, incremental:false)
+# so `killedBy` lists EVERY killer. Feed the uploaded reports to
+# `node scripts/quality/mutation-radiography.mjs --candidates mutation-nobail-*/mutation.json`
+# to get the accurate R1 prune-candidate list (🔴 empty ∪ 🟠 redundant) for human review.
+#
+# Batches mirror the nightly's leaf decomposition (d/e/f/g/h/i) rather than 2 mega-batches:
+# disableBail is MORE expensive than bail (it never stops early), and Stryker only writes
+# mutation.json on a SUCCESSFUL finish — a batch cancelled at the cap produces NO data — so
+# smaller batches each fit the 300min headroom and run in parallel. auth/accountFallback and
+# the security quartet are out of scope: R1 targets the combo/chatCore leaves.
+on:
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  stryker-nobail:
+    name: Stryker disableBail (batch ${{ matrix.batch.name }})
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        batch:
+          - name: d
+            mutate: "open-sse/services/combo/comboStructure.ts,open-sse/services/combo/autoStrategy.ts,open-sse/services/combo/validateQuality.ts"
+          - name: e
+            mutate: "open-sse/services/combo/shadowRouting.ts,open-sse/services/combo/targetSorters.ts,open-sse/services/combo/comboPredicates.ts,open-sse/services/combo/rrState.ts,open-sse/services/combo/comboData.ts"
+          - name: f
+            mutate: "open-sse/services/combo/quotaScoring.ts,open-sse/services/combo/quotaStrategies.ts"
+          - name: g
+            mutate: "open-sse/handlers/chatCore/comboContextCache.ts,open-sse/handlers/chatCore/idempotency.ts,open-sse/handlers/chatCore/passthroughHelpers.ts,open-sse/handlers/chatCore/responseHeaders.ts,open-sse/handlers/chatCore/sanitization.ts,open-sse/handlers/chatCore/upstreamTimeouts.ts"
+          - name: h
+            mutate: "open-sse/handlers/chatCore/headers.ts,open-sse/handlers/chatCore/logTruncation.ts,open-sse/handlers/chatCore/memoryExtraction.ts,open-sse/handlers/chatCore/nonStreamingSse.ts,open-sse/handlers/chatCore/passthroughToolNames.ts,open-sse/handlers/chatCore/executorHelpers.ts"
+          - name: i
+            mutate: "open-sse/handlers/chatCore/telemetryHelpers.ts,open-sse/handlers/chatCore/memorySkillsInjection.ts,open-sse/handlers/chatCore/semanticCache.ts"
+    timeout-minutes: 300
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: "24"
+          cache: npm
+      - run: npm ci
+      - name: Run Stryker (disableBail)
+        env:
+          BATCH_MUTATE: ${{ matrix.batch.mutate }}
+        run: npx stryker run --config-file stryker.disablebail.json --mutate "$BATCH_MUTATE"
+      - name: Upload mutation report
+        if: always()
+        uses: actions/upload-artifact@v7
+        with:
+          name: mutation-nobail-${{ matrix.batch.name }}
+          path: reports/mutation/
+          if-no-files-found: warn
+          retention-days: 14

.github/workflows/nightly-llm-security.yml

@@ -0,0 +1,102 @@
+name: Nightly LLM Security
+on:
+  schedule:
+    - cron: "53 5 * * *"
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  promptfoo-guard:
+    name: promptfoo — injection guard (block mode, no secret)
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with: { node-version: "24", cache: npm }
+      - run: npm ci
+      - name: Build CLI bundle
+        env: { JWT_SECRET: ci-build-secret-with-sufficient-length-for-validation }
+        run: npm run build:cli
+      - name: Start OmniRoute (block mode)
+        env:
+          JWT_SECRET: ci-build-secret-with-sufficient-length-for-validation
+          PORT: "20128"
+          INJECTION_GUARD_MODE: block
+        run: |
+          node dist/server.js > server.log 2>&1 &
+          echo $! > server.pid
+          for i in $(seq 1 30); do
+            if curl -sf http://localhost:20128/api/monitoring/health >/dev/null; then echo up; break; fi
+            sleep 2
+          done
+      - name: promptfoo guard-validation
+        run: npx --yes promptfoo@latest eval -c promptfooconfig.yaml --no-cache
+        env:
+          OMNIROUTE_URL: http://localhost:20128
+          OMNIROUTE_API_KEY: not-needed-blocked-before-upstream
+      - name: Stop server
+        if: always()
+        run: kill "$(cat server.pid)" || true
+
+  garak:
+    name: garak probes (skip without provider secret)
+    runs-on: ubuntu-latest
+    # NOTE: the `secrets` context is NOT available in a job-level `if:` — referencing
+    # it there makes GitHub reject the file on push (startup_failure on every push).
+    # Map the secret into a job-level env and gate each step on a presence check, so
+    # the job stays green and simply skips the probes when the secret is absent.
+    env:
+      PROMPTFOO_PROVIDER_KEY: ${{ secrets.PROMPTFOO_PROVIDER_KEY }}
+    steps:
+      - name: Gate on provider secret
+        id: gate
+        run: |
+          if [ -n "$PROMPTFOO_PROVIDER_KEY" ]; then
+            echo "run=true" >> "$GITHUB_OUTPUT"
+          else
+            echo "run=false" >> "$GITHUB_OUTPUT"
+            echo "::notice::PROMPTFOO_PROVIDER_KEY not set — skipping garak probes (advisory)."
+          fi
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+        if: steps.gate.outputs.run == 'true'
+      - uses: actions/setup-node@v6
+        if: steps.gate.outputs.run == 'true'
+        with: { node-version: "24", cache: npm }
+      - run: npm ci
+        if: steps.gate.outputs.run == 'true'
+      - name: Build CLI bundle
+        if: steps.gate.outputs.run == 'true'
+        env: { JWT_SECRET: ci-build-secret-with-sufficient-length-for-validation }
+        run: npm run build:cli
+      - name: Start OmniRoute
+        if: steps.gate.outputs.run == 'true'
+        env:
+          JWT_SECRET: ci-build-secret-with-sufficient-length-for-validation
+          PORT: "20128"
+        run: |
+          node dist/server.js > server.log 2>&1 &
+          echo $! > server.pid
+          for i in $(seq 1 30); do
+            if curl -sf http://localhost:20128/api/monitoring/health >/dev/null; then echo up; break; fi
+            sleep 2
+          done
+      - uses: actions/setup-python@v6
+        if: steps.gate.outputs.run == 'true'
+        with: { python-version: "3.12" }
+      - run: pip install garak
+        if: steps.gate.outputs.run == 'true'
+      - name: garak limited probes
+        if: steps.gate.outputs.run == 'true'
+        env:
+          OPENAI_API_KEY: ${{ secrets.PROMPTFOO_PROVIDER_KEY }}
+          OPENAI_BASE_URL: http://localhost:20128/v1
+        run: garak --model_type openai --model_name gpt-4o-mini --probes promptinject,dan,leakreplay --report_prefix garak-omniroute || true
+      - name: Stop server
+        if: always() && steps.gate.outputs.run == 'true'
+        run: kill "$(cat server.pid)" || true

.github/workflows/nightly-mutation.yml

@@ -0,0 +1,163 @@
+name: Nightly Mutation
+on:
+  schedule:
+    - cron: "17 3 * * *"
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  stryker:
+    name: Stryker mutation (batch ${{ matrix.batch.name }} — advisory)
+    runs-on: ubuntu-latest
+    # Mutation testing is expensive. History of the budget:
+    #   - Full 8-module set TIMED OUT at the 180min cap (run 27705123780 = exactly 180min).
+    #     The two god-files chatCore.ts/combo.ts dominated ~2/3 of the mutants and were
+    #     removed from stryker.conf.json `mutate`.
+    #   - The remaining 6 modules in 3 batches: auth.ts and accountFallback.ts are large; the
+    #     a (auth+publicCreds) and b (accountFallback+error) batches still ran near the 180min
+    #     cap (the perTest dry-run over ~130 covering test files is itself costly), so the two
+    #     big modules are now ISOLATED into their own batches (a=auth, b=accountFallback).
+    #   - Onda 3 / Fase 9 T5 re-add: combo.ts was split into 11 leaves; the 8 well-covered
+    #     combo/* leaves are back in `mutate` (covered by the 24 combo-*.test.ts), grouped into
+    #     2 batches (d=heavy, e=light). After #4204 (D7b) merged, the reset-aware quota pair
+    #     quotaScoring/quotaStrategies was added as batch f. A covering-test audit then added the
+    #     6 chatCore/* leaves with direct unit coverage as batch g. A follow-up then wrote dedicated
+    #     unit tests for 6 more leaves and added them as batch h. A final follow-up added dedicated
+    #     tests (no mock.module — fetch-override + crafted inputs + temp-DATA_DIR) for telemetryHelpers
+    #     + memorySkillsInjection + semanticCache (its cache-HIT block now has a setCachedResponse
+    #     fixture) as batch i — ALL 15/15 chatCore leaves are now mutated. See
+    #     _mutate_godfiles_excluded_comment in stryker.conf.json.
+    # 9 PARALLEL batches, each overriding the mutate set via `--mutate` (Stryker 9 CLI:
+    # `-m, --mutate <comma-list>`; the conf's `mutate[]` remains the local-run default/union).
+    #   - Cold-seeding budget (per-batch `timeout-minutes: ${{ matrix.batch.timeout || 180 }}`):
+    #     a COLD run must COMPLETE once to write stryker-incremental.json (Stryker writes it only on
+    #     a successful finish); a job cancelled at the cap writes nothing, so the next run is cold
+    #     again — an infinite never-seeds loop. actions/cache is also branch-scoped, so each branch
+    #     (incl. release) must seed its OWN cache via a run with enough headroom. Measured cold runs
+    #     Measured cold totals (run 27801802713, extrapolated from the % at the 180/350 cancel point):
+    #     auth ~375min (2301 mutants — EXCEEDS the 360min job max even isolated), accountFallback
+    #     ~358min (1441), the c security quartet ~348min (1163), d ~197min (1316), g=142, h=132, e=66,
+    #     f=45, i=33. The widely-covered modules blow the budget because the tap-runner re-runs every
+    #     covering test file per mutant (a perTest fixed cost over ~138 test files, times thousands of
+    #     mutants). A flat timeout bump cannot rescue auth (>360min max) — so the three over-budget
+    #     batches are SPLIT so each half fits: auth->a1/a2 and accountFallback->b1/b2 by mutation range
+    #     (`file:startLine-endLine`), the c quartet->c1/c2 by module pair. Splitting also seeds each
+    #     sub-batch's own incremental cache, after which nightlies re-test only changed mutants.
+    # Full coverage every night in parallel; wall-clock = the slowest batch's cold run until seeded.
+    # Runs at stryker concurrency=4 with per-process DATA_DIR isolation
+    # (tests/_setup/isolateDataDir.ts) — see _concurrency_comment in stryker.conf.json.
+    strategy:
+      fail-fast: false
+      matrix:
+        batch:
+          # Per-batch `timeout` (minutes) tiers the cold-seeding budget by measured cost; batches
+          # without the key default to 180. Cold-run profiling (run 27801802713) showed the
+          # widely-covered modules need FAR more than the 180 cap because the tap-runner re-runs every
+          # covering test file per mutant: auth ~375min (2301 mutants, EXCEEDS the 360min GitHub job
+          # max even isolated), accountFallback ~358min, the c security quartet ~348min — none fit a
+          # single job. So auth/accountFallback are split by MUTATION RANGE (`file:startLine-endLine`,
+          # ~half the mutants each) into a1/a2, b1/b2; the c quartet is split by MODULE pair into
+          # c1/c2. d (3 combo modules, ~197min) stays whole. Split-heavy batches get 300min headroom
+          # for their cold seeding run; once each batch completes once and writes
+          # stryker-incremental.json, later runs re-test only changed mutants and finish far faster.
+          # g/h (142/132min cold) keep a 240 buffer; e/f/i (33-66min) keep the 180 default.
+          - name: a1
+            mutate: "src/sse/services/auth.ts:1-1109"
+            timeout: 300
+          - name: a2
+            mutate: "src/sse/services/auth.ts:1110-2218"
+            timeout: 300
+          - name: b1
+            mutate: "open-sse/services/accountFallback.ts:1-863"
+            timeout: 300
+          - name: b2
+            mutate: "open-sse/services/accountFallback.ts:864-1726"
+            timeout: 300
+          - name: c1
+            mutate: "src/server/authz/routeGuard.ts,src/shared/utils/circuitBreaker.ts"
+            timeout: 300
+          - name: c2
+            mutate: "open-sse/utils/error.ts,open-sse/utils/publicCreds.ts"
+            timeout: 300
+          - name: d
+            mutate: "open-sse/services/combo/comboStructure.ts,open-sse/services/combo/autoStrategy.ts,open-sse/services/combo/validateQuality.ts"
+            timeout: 300
+          - name: e
+            mutate: "open-sse/services/combo/shadowRouting.ts,open-sse/services/combo/targetSorters.ts,open-sse/services/combo/comboPredicates.ts,open-sse/services/combo/rrState.ts,open-sse/services/combo/comboData.ts"
+          - name: f
+            mutate: "open-sse/services/combo/quotaScoring.ts,open-sse/services/combo/quotaStrategies.ts"
+          - name: g
+            mutate: "open-sse/handlers/chatCore/comboContextCache.ts,open-sse/handlers/chatCore/idempotency.ts,open-sse/handlers/chatCore/passthroughHelpers.ts,open-sse/handlers/chatCore/responseHeaders.ts,open-sse/handlers/chatCore/sanitization.ts,open-sse/handlers/chatCore/upstreamTimeouts.ts"
+            timeout: 240
+          - name: h
+            mutate: "open-sse/handlers/chatCore/headers.ts,open-sse/handlers/chatCore/logTruncation.ts,open-sse/handlers/chatCore/memoryExtraction.ts,open-sse/handlers/chatCore/nonStreamingSse.ts,open-sse/handlers/chatCore/passthroughToolNames.ts,open-sse/handlers/chatCore/executorHelpers.ts"
+            timeout: 240
+          - name: i
+            mutate: "open-sse/handlers/chatCore/telemetryHelpers.ts,open-sse/handlers/chatCore/memorySkillsInjection.ts,open-sse/handlers/chatCore/semanticCache.ts"
+    # Per-batch budget: split-heavy batches (a1/a2/b1/b2/c1/c2/d) override to 300min, g/h to 240min;
+    # the rest default to 180min. `matrix.batch.timeout` is null for batches without the key -> `|| 180`.
+    # NOTE: a1+a2 both mutate auth.ts (disjoint line ranges) and b1+b2 both mutate accountFallback.ts;
+    # when merging the per-batch mutation.json for radiography/scores, same-file mutants from sibling
+    # ranges must be UNIONED (scripts/check/check-mutation-ratchet.mjs::measureMutationScores and
+    # scripts/quality/mutation-radiography.mjs both merge per file).
+    timeout-minutes: ${{ matrix.batch.timeout || 180 }}
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: "24"
+          cache: npm
+      - run: npm ci
+      - name: Restore Stryker incremental cache
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae # v5.0.5
+        with:
+          path: reports/mutation/stryker-incremental.json
+          key: stryker-incremental-${{ matrix.batch.name }}-${{ github.run_id }}
+          restore-keys: stryker-incremental-${{ matrix.batch.name }}-
+      - name: Run Stryker (advisory)
+        id: stryker
+        continue-on-error: true
+        env:
+          BATCH_MUTATE: ${{ matrix.batch.mutate }}
+        run: npx stryker run --mutate "$BATCH_MUTATE"
+      - name: Upload mutation report
+        if: always()
+        uses: actions/upload-artifact@v7
+        with:
+          name: mutation-report-${{ matrix.batch.name }}
+          path: reports/mutation/
+          if-no-files-found: warn
+          retention-days: 14
+
+  # Aggregation gate (T3): each split batch emits a PARTIAL view of a mutated file
+  # (auth.ts lives in a1+a2, accountFallback in b1+b2), so a PER-BATCH ratchet would
+  # only ever see half a file vs the whole-file baseline. This job runs AFTER every
+  # batch, downloads all reports, and ratchets the MERGED per-module scores
+  # (check-mutation-ratchet UNIONS same-file mutants across reports) against the
+  # dedicatedGate `mutationScore.*` floors in quality-baseline.json (seeded ~2pt below
+  # the first full measurement). Blocking: a module dropping below its floor fails the
+  # run. Missing reports (e.g. an artifact-upload flake) are skipped, never failed.
+  mutation-ratchet:
+    name: Mutation score ratchet (blocking)
+    needs: stryker
+    if: always()
+    runs-on: ubuntu-latest
+    timeout-minutes: 15
+    steps:
+      - uses: actions/checkout@v6
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: "24"
+      - name: Download all mutation reports
+        uses: actions/download-artifact@v8
+        with:
+          pattern: mutation-report-*
+          path: reports/all
+      - name: Ratchet merged per-module mutation scores
+        run: node scripts/check/check-mutation-ratchet.mjs reports/all/*/mutation.json --ratchet

.github/workflows/nightly-property.yml

@@ -0,0 +1,35 @@
+name: Nightly Property Discovery
+on:
+  schedule:
+    - cron: "0 6 * * *"
+  workflow_dispatch:
+permissions:
+  contents: read
+  issues: write
+jobs:
+  property-random-seed:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: "24"
+          cache: npm
+      - run: npm ci
+      - name: fast-check random seed (high runs)
+        id: prop
+        run: FC_SEED=random FC_NUM_RUNS=2000 npm run test:property
+      - name: Open issue on failure
+        if: failure()
+        uses: actions/github-script@v9
+        with:
+          script: |
+            await github.rest.issues.create({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              title: "Nightly property-test failure (Fase 8 B)",
+              body: "fast-check found a counterexample with a random seed. Check the run logs for the reproducible seed + minimal case, then add it as a fixture.\n\nRun: " + context.serverUrl + "/" + context.repo.owner + "/" + context.repo.repo + "/actions/runs/" + context.runId,
+              labels: ["quality-gate-finding"],
+            });

.github/workflows/nightly-release-green.yml

@@ -0,0 +1,137 @@
+name: Nightly Release-Green
+
+# Solution D — continuous, NON-BLOCKING drift signal for the active release branch.
+#
+# WHY: the full gate (ci.yml) only runs on the release PR (PR → main), so reds
+# accrue silently on release/** and explode — in layers — at release time. This
+# nightly reproduces the release-equivalent validation on the active release branch
+# HEAD and, when there are HARD failures, opens/updates a single tracking issue.
+#
+# It is NOT a required status check and never touches a contributor PR — it only
+# reports. Ratchet drift (eslint warnings / cognitive-complexity / file-size) is
+# expected mid-cycle and is reported but never raises the alarm on its own; only
+# real defects (typecheck / lint errors / unit / vitest / db-rules / public-creds /
+# package-artifact) flip the issue open.
+
+on:
+  schedule:
+    - cron: "23 5 * * *" # 05:23 UTC daily — off-peak, distinct from other nightlies
+  workflow_dispatch:
+    inputs:
+      branch:
+        description: "Release branch to validate (default: highest release/vX.Y.Z)"
+        required: false
+        type: string
+
+permissions:
+  contents: read
+  issues: write
+
+concurrency:
+  group: nightly-release-green
+  cancel-in-progress: true
+
+jobs:
+  release-green:
+    name: Validate active release branch
+    runs-on: ubuntu-latest
+    env:
+      JWT_SECRET: ci-nightly-secret-with-sufficient-length-for-validation
+      API_KEY_SECRET: ci-nightly-api-key-secret-long
+      DISABLE_SQLITE_AUTO_BACKUP: "true"
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          fetch-depth: 0
+          persist-credentials: false
+
+      - name: Resolve active release branch
+        id: branch
+        env:
+          INPUT_BRANCH: ${{ github.event.inputs.branch }}
+        run: |
+          set -euo pipefail
+          if [ -n "${INPUT_BRANCH:-}" ]; then
+            TARGET="$INPUT_BRANCH"
+          else
+            # highest release/vX.Y.Z by semver among remote branches
+            TARGET=$(git for-each-ref --format='%(refname:short)' 'refs/remotes/origin/release/v*' \
+              | sed 's#origin/##' \
+              | sort -t/ -k2 -V \
+              | tail -1)
+          fi
+          if [ -z "$TARGET" ]; then echo "No release/v* branch found"; exit 1; fi
+          # Strict format guard — reject anything that isn't release/vX.Y.Z (blocks
+          # ref/command injection via the workflow_dispatch input).
+          if ! printf '%s' "$TARGET" | grep -qE '^release/v[0-9]+\.[0-9]+\.[0-9]+$'; then
+            echo "Refusing non-canonical branch name: $TARGET"; exit 1
+          fi
+          echo "target=$TARGET" >> "$GITHUB_OUTPUT"
+          echo "Active release branch: $TARGET"
+
+      - name: Checkout the release branch
+        env:
+          TARGET: ${{ steps.branch.outputs.target }}
+        run: |
+          set -euo pipefail
+          git checkout "$TARGET"
+          git log -1 --oneline
+
+      - uses: actions/setup-node@v6
+        with:
+          node-version: "24"
+          cache: npm
+
+      - uses: ./.github/actions/npm-ci-retry
+
+      - name: Release-green validation (full)
+        id: validate
+        run: |
+          set +e
+          node scripts/quality/validate-release-green.mjs --json --with-build \
+            1> release-green.json 2> release-green.log
+          echo "exit=$?" >> "$GITHUB_OUTPUT"
+          echo "------- report -------"
+          cat release-green.log
+
+      - name: Open / update tracking issue on HARD failure
+        if: steps.validate.outputs.exit != '0'
+        env:
+          GH_TOKEN: ${{ github.token }}
+          TARGET: ${{ steps.branch.outputs.target }}
+          RUN_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
+        run: |
+          set -euo pipefail
+          TITLE="🔴 Release branch not green: ${TARGET}"
+          {
+            echo "The nightly **release-green** validation found HARD failures on \`${TARGET}\`."
+            echo "These are real defects that would block the release PR — fix them in the"
+            echo "originating PR branch (via co-authorship), not by demanding it from contributors."
+            echo ""
+            echo "**Run:** ${RUN_URL}"
+            echo ""
+            echo '```'
+            sed -n '/──────── verdict ────────/,$p' release-green.log || tail -40 release-green.log
+            echo '```'
+            echo ""
+            echo "_Ratchet drift (eslint warnings / cognitive-complexity / file-size) listed above is expected mid-cycle and is rebaselined at release — it is NOT a contributor concern and did not, on its own, open this issue._"
+          } > issue-body.md
+
+          EXISTING=$(gh issue list --repo "$GITHUB_REPOSITORY" --state open \
+            --search "in:title $TITLE" --json number --jq '.[0].number' 2>/dev/null || echo "")
+          if [ -n "$EXISTING" ]; then
+            gh issue comment "$EXISTING" --repo "$GITHUB_REPOSITORY" --body-file issue-body.md
+            echo "Updated existing issue #$EXISTING"
+          else
+            gh issue create --repo "$GITHUB_REPOSITORY" --title "$TITLE" --body-file issue-body.md
+          fi
+
+      - name: Upload report artifact
+        if: always()
+        uses: actions/upload-artifact@v4
+        with:
+          name: release-green-report
+          path: |
+            release-green.json
+            release-green.log
+          if-no-files-found: ignore

.github/workflows/nightly-resilience.yml

@@ -0,0 +1,110 @@
+name: Nightly Resilience
+on:
+  schedule:
+    - cron: "41 4 * * *"
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  heap:
+    name: Heap-growth gate
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: "24"
+          cache: npm
+      - run: npm ci
+      - run: npm run test:heap
+
+  chaos:
+    name: Resilience chaos (fault injection)
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: "24"
+          cache: npm
+      - run: npm ci
+      - run: npm run test:chaos
+
+  k6-soak:
+    name: k6 load/soak
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: "24"
+          cache: npm
+      - run: npm ci
+      - name: Build CLI bundle
+        env:
+          JWT_SECRET: ci-build-secret-with-sufficient-length-for-validation
+        run: npm run build:cli
+      - name: Start OmniRoute (background)
+        env:
+          JWT_SECRET: ci-build-secret-with-sufficient-length-for-validation
+          PORT: "20128"
+        run: |
+          node dist/server.js > server.log 2>&1 &
+          echo $! > server.pid
+          for i in $(seq 1 30); do
+            if curl -sf http://localhost:20128/api/monitoring/health >/dev/null; then echo "server up"; break; fi
+            sleep 2
+          done
+      - name: Install k6
+        uses: grafana/setup-k6-action@v1
+      - name: Run k6 soak
+        run: k6 run tests/load/k6-soak.js
+        env:
+          BASE_URL: http://localhost:20128
+          SOAK_DURATION: "3m"
+          SOAK_VUS: "10"
+      - name: Stop server
+        if: always()
+        run: kill "$(cat server.pid)" || true
+
+  a11y:
+    name: A11y axe (nightly, freeze-and-alert)
+    runs-on: ubuntu-latest
+    # The Playwright webServer (`start` mode) builds Next via build-next-isolated.mjs and
+    # boots the standalone server itself (waits on /api/monitoring/health, 15min webServer
+    # timeout). Unlike the per-PR test-e2e job, this nightly job has no pre-built artifact,
+    # so it self-builds — hence the generous job timeout. REQUIRE_AXE=1 makes the suite run
+    # the real axe analysis (the 4 page tests are gated to nightly so per-PR e2e stays fast)
+    # and makes the meta-test fail loudly if @axe-core/playwright ever goes missing.
+    timeout-minutes: 30
+    env:
+      JWT_SECRET: ci-test-secret-with-sufficient-length-for-validation
+      API_KEY_SECRET: ci-test-api-key-secret-long
+      DISABLE_SQLITE_AUTO_BACKUP: "true"
+      REQUIRE_AXE: "1"
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: "24"
+          cache: npm
+      - run: npm ci
+      - name: Cache Playwright browsers
+        uses: actions/cache@v5.0.5
+        with:
+          path: ~/.cache/ms-playwright
+          key: playwright-chromium-${{ runner.os }}-${{ hashFiles('package-lock.json') }}
+          restore-keys: playwright-chromium-${{ runner.os }}-
+      - run: npx playwright install --with-deps chromium
+      - name: Run axe a11y suite (self-building webServer)
+        run: npx playwright test tests/e2e/a11y.spec.ts

.github/workflows/nightly-schemathesis.yml

@@ -0,0 +1,72 @@
+name: Nightly Schemathesis
+on:
+  schedule:
+    - cron: "23 4 * * *"
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  schemathesis:
+    name: Schemathesis — OpenAPI contract fuzz (advisory)
+    runs-on: ubuntu-latest
+    timeout-minutes: 30
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with: { node-version: "24", cache: npm }
+      - run: npm ci
+      - name: Build CLI bundle
+        env: { JWT_SECRET: ci-build-secret-with-sufficient-length-for-validation }
+        run: npm run build:cli
+      - name: Start OmniRoute (background)
+        env:
+          JWT_SECRET: ci-build-secret-with-sufficient-length-for-validation
+          PORT: "20128"
+        run: |
+          node dist/server.js > server.log 2>&1 &
+          echo $! > server.pid
+          for i in $(seq 1 30); do
+            if curl -sf http://localhost:20128/api/monitoring/health >/dev/null; then echo "server up"; break; fi
+            sleep 2
+          done
+      - uses: actions/setup-python@v6
+        with: { python-version: "3.12" }
+      - name: Install schemathesis
+        run: pip install schemathesis
+      - name: Schemathesis contract fuzz (advisory)
+        # Advisory gate: never fails the job. `continue-on-error` covers a crash of the
+        # step itself; `|| true` covers schemathesis exiting non-zero when it finds spec
+        # violations / upstream 500s — both are expected here (most /v1 endpoints proxy an
+        # upstream that has no provider configured in CI). The point of the nightly is to
+        # PROVE the contract is fuzzable and surface regressions, not to gate the build.
+        continue-on-error: true
+        run: |
+          schemathesis run docs/openapi.yaml \
+            --url http://localhost:20128 \
+            --max-examples 20 \
+            --workers 4 \
+            --checks all \
+            --max-response-time 30 \
+            --request-timeout 30 \
+            --suppress-health-check all \
+            --report junit \
+            --report-junit-path schemathesis-report/junit.xml \
+            --no-color \
+            || true
+      - name: Stop server
+        if: always()
+        run: kill "$(cat server.pid)" || true
+      - name: Upload schemathesis report
+        if: always()
+        uses: actions/upload-artifact@v7
+        with:
+          name: schemathesis-report
+          path: |
+            schemathesis-report/
+            server.log
+          if-no-files-found: warn
+          retention-days: 14

.github/workflows/npm-publish.yml

@@ -1,42 +1,242 @@
 name: Publish to npm
 
 on:
+  # 'released' (not 'published') so editing/re-publishing old releases does NOT
+  # re-trigger this workflow. Pairs with the semver guard below as defense in
+  # depth against accidental dist-tag clobbering by old releases.
   release:
-    types: [published]
+    types: [released]
+  workflow_dispatch:
+    inputs:
+      version:
+        description: "Version to publish (e.g. 2.9.5 or 3.0.0-rc.15)"
+        required: true
+        type: string
+      tag:
+        description: "npm dist-tag (auto / latest / next / historic)"
+        required: false
+        default: "auto"
+        type: choice
+        options:
+          - auto
+          - latest
+          - next
+          - historic
+  workflow_call:
+    inputs:
+      version:
+        description: "Version to publish (without v prefix)"
+        required: true
+        type: string
+      tag:
+        description: "npm dist-tag (auto / latest / next / historic)"
+        required: false
+        default: "auto"
+        type: string
+    secrets:
+      NPM_TOKEN:
+        required: true
 
+# Least-privilege default: read-only at the top level; each publish job grants the
+# id-token (npm provenance) / packages (GitHub Packages) writes it needs (Scorecard
+# TokenPermissions).
 permissions:
   contents: read
-  id-token: write
+
+env:
+  NPM_PUBLISH_NODE_VERSION: "24"
 
 jobs:
   publish:
     runs-on: ubuntu-latest
-    environment: NPM_TOKEN
+    permissions:
+      contents: write # gh release upload (attach SBOM to the GitHub Release)
+      id-token: write # npm provenance
+      packages: write # publish to npm.pkg.github.com
     steps:
       - name: Checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+          # Need full tag history to compare against highest semver when
+          # deciding whether this release should claim dist-tag `latest`.
+          fetch-depth: 0
 
       - name: Setup Node.js
         uses: actions/setup-node@v6
         with:
-          node-version: 22
+          node-version: ${{ env.NPM_PUBLISH_NODE_VERSION }}
           registry-url: https://registry.npmjs.org
 
-      - name: Install dependencies
-        run: npm ci
+      - name: Install dependencies (skip scripts to avoid heavy build)
+        run: npm install --ignore-scripts --no-audit --no-fund
 
-      - name: Build standalone app
+      - name: Resolve version, dist-tag and skip flag
+        id: resolve
+        env:
+          EVENT_NAME: ${{ github.event_name }}
+          REF_NAME: ${{ github.ref_name }}
+          INPUT_VERSION: ${{ inputs.version }}
+          INPUT_TAG: ${{ inputs.tag }}
+        run: |
+          set -euo pipefail
+
+          # 1) Resolve VERSION from the trigger (all inputs come via env).
+          VERSION="${INPUT_VERSION:-}"
+          if [ -z "$VERSION" ] && [ "$EVENT_NAME" = "release" ]; then
+            VERSION="$REF_NAME"
+          fi
+          VERSION="${VERSION#v}"
+          if ! printf '%s' "$VERSION" | grep -qE '^[0-9]+\.[0-9]+\.[0-9]+([.-][A-Za-z0-9.-]+)?$'; then
+            echo "Refusing to publish unsafe VERSION value: $VERSION" >&2
+            exit 1
+          fi
+
+          # 2) Resolve dist-tag.
+          #    - explicit 'latest'/'next'/'historic' is honored
+          #    - 'auto' (or empty): pre-release identifiers → 'next';
+          #      stable versions → 'latest' only if VERSION is the highest
+          #      stable semver among `v*` tags (otherwise → 'historic').
+          REQUESTED_TAG="${INPUT_TAG:-auto}"
+          TAG="$REQUESTED_TAG"
+          if [ "$TAG" = "auto" ] || [ -z "$TAG" ]; then
+            if printf '%s' "$VERSION" | grep -qE -- '-(rc|alpha|beta|pre|next)'; then
+              TAG="next"
+            else
+              git fetch --tags --quiet || true
+              HIGHEST=$(git tag -l 'v[0-9]*' | sed 's/^v//' | grep -vE -- '-(rc|alpha|beta|pre|next)' | sort -V | tail -1 || echo "")
+              if [ -n "$HIGHEST" ] && [ "$VERSION" = "$HIGHEST" ]; then
+                TAG="latest"
+              else
+                echo "Version $VERSION is not the highest semver tag (highest=${HIGHEST:-<none>}). Using dist-tag 'historic' to avoid clobbering @latest."
+                TAG="historic"
+              fi
+            fi
+          fi
+
+          # 3) Skip-if-already-published. NOTE: do NOT pass `--silent` to
+          #    `npm view` — it suppresses stdout and breaks the grep, which
+          #    caused old releases (3.2.8) to be re-published and steal
+          #    dist-tag `latest`. See incident notes in CHANGELOG.
+          PUBLISHED="$(npm view "omniroute@${VERSION}" version 2>/dev/null || true)"
+          SKIP="false"
+          if [ "$PUBLISHED" = "$VERSION" ]; then
+            echo "⚠️ omniroute@${VERSION} is already on npm — skipping publish."
+            SKIP="true"
+          fi
+
+          echo "version=$VERSION" >> "$GITHUB_OUTPUT"
+          echo "tag=$TAG"          >> "$GITHUB_OUTPUT"
+          echo "skip=$SKIP"        >> "$GITHUB_OUTPUT"
+          echo "📦 Resolved omniroute@$VERSION dist-tag=$TAG skip=$SKIP"
+
+      - name: Sync package.json version
+        if: steps.resolve.outputs.skip != 'true'
+        env:
+          VERSION: ${{ steps.resolve.outputs.version }}
+        run: |
+          npm version "$VERSION" --no-git-tag-version --allow-same-version
+
+      - name: Build CLI bundle (standalone app)
+        if: steps.resolve.outputs.skip != 'true'
+        env:
+          JWT_SECRET: ci-build-secret-with-sufficient-length-for-validation
         run: npm run build:cli
 
-      - name: Sync version from release tag
-        run: |
-          VERSION="${GITHUB_REF_NAME}"
-          # Remove 'v' prefix if present (v0.1.0 -> 0.1.0)
-          VERSION="${VERSION#v}"
-          npm version "$VERSION" --no-git-tag-version --allow-same-version
-          echo "Publishing version: $VERSION"
+      - name: Validate npm package artifact
+        if: steps.resolve.outputs.skip != 'true'
+        run: npm run check:pack-artifact
+
+      - name: Generate CycloneDX SBOM (npm)
+        if: steps.resolve.outputs.skip != 'true'
+        run: npx @cyclonedx/cyclonedx-npm --ignore-npm-errors --output-format JSON --output-file sbom-npm.cdx.json
+
+      - name: Upload SBOM (npm) as workflow artifact
+        if: steps.resolve.outputs.skip != 'true'
+        uses: actions/upload-artifact@v7
+        with:
+          name: sbom-npm
+          path: sbom-npm.cdx.json
+          if-no-files-found: error
+
+      - name: Attach SBOM to GitHub Release
+        if: steps.resolve.outputs.skip != 'true' && github.event_name == 'release'
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          TAG: ${{ github.ref_name }}
+        run: gh release upload "$TAG" sbom-npm.cdx.json --clobber
 
       - name: Publish to npm
-        run: npm publish --access public
+        if: steps.resolve.outputs.skip != 'true'
+        env:
+          VERSION: ${{ steps.resolve.outputs.version }}
+          TAG: ${{ steps.resolve.outputs.tag }}
+          NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
+        run: |
+          set -euo pipefail
+          # Always pass --tag explicitly. Defense in depth: even if VERSION is
+          # accidentally an older release, `npm publish --tag historic` will
+          # NOT promote it to `@latest`.
+          npm publish --provenance --access public --tag "$TAG"
+          echo "✅ Published omniroute@$VERSION (dist-tag=$TAG)"
+
+      - name: Publish to GitHub Packages
+        if: steps.resolve.outputs.skip != 'true'
+        env:
+          VERSION: ${{ steps.resolve.outputs.version }}
+          TAG: ${{ steps.resolve.outputs.tag }}
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: |
+          set -euo pipefail
+          echo "Configuring for GitHub Packages..."
+          echo "//npm.pkg.github.com/:_authToken=${GITHUB_TOKEN}" > .npmrc
+          npm pkg set name="@diegosouzapw/omniroute"
+          npm publish --registry=https://npm.pkg.github.com --tag "$TAG" \
+            || echo "⚠️ omniroute@${VERSION} might already be published on GitHub Packages."
+          echo "✅ Action finished for GitHub Packages"
+
+  publish-opencode-plugin:
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      id-token: write # npm provenance
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+
+      - name: Setup Node.js
+        uses: actions/setup-node@v6
+        with:
+          node-version: ${{ env.NPM_PUBLISH_NODE_VERSION }}
+          registry-url: https://registry.npmjs.org
+
+      - name: Install plugin dependencies
+        working-directory: "@omniroute/opencode-plugin"
+        run: npm install --no-audit --no-fund
+
+      - name: Build plugin
+        working-directory: "@omniroute/opencode-plugin"
+        run: npm run clean && npm run build
+
+      - name: Test plugin
+        working-directory: "@omniroute/opencode-plugin"
+        run: npm test
+
+      - name: Publish @omniroute/opencode-plugin to npm
+        working-directory: "@omniroute/opencode-plugin"
         env:
           NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
+        run: |
+          set -euo pipefail
+          PKG_VERSION=$(node -p "require('./package.json').version")
+          PKG_NAME=$(node -p "require('./package.json').name")
+          # Same hardened skip-check as the main job (no --silent flag).
+          PUBLISHED="$(npm view "${PKG_NAME}@${PKG_VERSION}" version 2>/dev/null || true)"
+          if [ "$PUBLISHED" = "$PKG_VERSION" ]; then
+            echo "⚠️ ${PKG_NAME}@${PKG_VERSION} is already published on npm — skipping."
+            exit 0
+          fi
+          npm publish --provenance --access public --ignore-scripts
+          echo "✅ Published ${PKG_NAME}@${PKG_VERSION}"

.github/workflows/opencode-plugin-ci.yml

@@ -0,0 +1,66 @@
+name: opencode-plugin CI
+
+on:
+  push:
+    branches: [main, release/v3.8.2]
+    paths:
+      - "@omniroute/opencode-plugin/**"
+  pull_request:
+    branches: [main, release/v3.8.2]
+    paths:
+      - "@omniroute/opencode-plugin/**"
+    types: [opened, synchronize, reopened, ready_for_review]
+  workflow_dispatch:
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+permissions:
+  contents: read
+
+defaults:
+  run:
+    working-directory: "@omniroute/opencode-plugin"
+
+jobs:
+  test:
+    name: Test (Node ${{ matrix.node }})
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        node: ["22", "24"]
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: ${{ matrix.node }}
+          cache: npm
+          cache-dependency-path: "@omniroute/opencode-plugin/package-lock.json"
+      - run: npm install --no-audit --no-fund
+      - run: npm run build
+      - run: npm test
+
+  build:
+    name: Build
+    runs-on: ubuntu-latest
+    needs: test
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: "22"
+          cache: npm
+          cache-dependency-path: "@omniroute/opencode-plugin/package-lock.json"
+      - run: npm install --no-audit --no-fund
+      - run: npm run build
+      - uses: actions/upload-artifact@v7
+        with:
+          name: opencode-plugin-dist
+          path: "@omniroute/opencode-plugin/dist"
+          retention-days: 7

.github/workflows/opencode-provider-ci.yml

@@ -0,0 +1,65 @@
+name: opencode-provider CI
+
+on:
+  push:
+    branches: [main, release/v3.8.0]
+    paths:
+      - "@omniroute/opencode-provider/**"
+  pull_request:
+    branches: [main, release/v3.8.0]
+    paths:
+      - "@omniroute/opencode-provider/**"
+    types: [opened, synchronize, reopened, ready_for_review]
+  workflow_dispatch:
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+permissions:
+  contents: read
+
+defaults:
+  run:
+    working-directory: "@omniroute/opencode-provider"
+
+jobs:
+  test:
+    name: Test (Node ${{ matrix.node }})
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        node: ["20", "22", "24"]
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: ${{ matrix.node }}
+          cache: npm
+          cache-dependency-path: "@omniroute/opencode-provider/package-lock.json"
+      - run: npm ci
+      - run: npm test
+
+  build:
+    name: Build
+    runs-on: ubuntu-latest
+    needs: test
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: "20"
+          cache: npm
+          cache-dependency-path: "@omniroute/opencode-provider/package-lock.json"
+      - run: npm ci
+      - run: npm run build
+      - uses: actions/upload-artifact@v7
+        with:
+          name: opencode-provider-dist
+          path: "@omniroute/opencode-provider/dist"
+          retention-days: 7

.github/workflows/quality.yml

@@ -0,0 +1,84 @@
+name: Quality Gates
+
+on:
+  pull_request:
+    branches: ["release/**"]
+    types: [opened, synchronize, reopened, ready_for_review]
+  workflow_dispatch:
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+
+permissions:
+  contents: read
+
+env:
+  CI_NODE_VERSION: "24"
+
+jobs:
+  fast-gates:
+    name: Fast Quality Gates
+    runs-on: ubuntu-latest
+    # tsx gates (known-symbols, route-guard-membership) import modules that open
+    # SQLite on load; provide DB env so a fresh CI DB initializes cleanly.
+    env:
+      JWT_SECRET: ci-lint-secret-with-sufficient-length-for-validation
+      API_KEY_SECRET: ci-lint-api-key-secret-long
+      DISABLE_SQLITE_AUTO_BACKUP: "true"
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          fetch-depth: 0
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: ${{ env.CI_NODE_VERSION }}
+          cache: npm
+      - run: npm ci
+      - run: npm run check:provider-consistency
+      - run: npm run check:fetch-targets
+      - run: npm run check:openapi-routes
+      - run: npm run check:docs-symbols
+      - run: npm run check:deps
+      - run: npm run check:file-size
+      - run: npm run check:error-helper
+      - run: npm run check:migration-numbering
+      - run: npm run check:public-creds
+      - run: npm run check:db-rules
+      - run: npm run check:known-symbols
+      - run: npm run check:route-guard-membership
+      - run: npm run check:test-discovery
+      - run: npm run check:any-budget:t11
+      - name: Typecheck (core)
+        run: npm run typecheck:core
+      # TIA: build the impact map at runtime (gitignored, ~21MB) and run only the
+      # unit tests impacted by this PR's changed files. Fail-safe runs the FULL
+      # unit suite on hub/unmapped changes — TIA accelerates, never replaces, the net.
+      #
+      # BLOCKING (flipped 2026-06-17). The pre-existing release unit test-debt that kept
+      # this advisory was cleared: #4030 (16 Zod/registry reds, lossless restore) and
+      # #4063 (the last red — the LiveWS boot test — root-caused as a real event-loop
+      # stall in the WS sidecar, fixed + relocated to the integration suite). A full
+      # ci.yml run on release/v3.8.28 then showed all 8 unit shards green, so PR->release
+      # now blocks on unit-test regressions in the impacted set (typecheck:core already
+      # blocked above). Fail-safe still runs the FULL unit suite on hub/unmapped changes.
+      - name: Impacted unit tests (TIA, fail-safe full; blocking)
+        env:
+          GITHUB_BASE_REF: ${{ github.base_ref }}
+        run: |
+          git fetch --no-tags origin "$GITHUB_BASE_REF" || true
+          node scripts/quality/build-test-impact-map.mjs
+          SEL="$(node scripts/quality/select-impacted-tests.mjs)"
+          if [ -z "$SEL" ]; then echo "No source/test changes — skipping unit tests"; exit 0; fi
+          # CI runners are 4-vCPU; run at --test-concurrency=4 (matching the ci.yml unit
+          # job) rather than test:unit's local-tuned concurrency=20. Oversubscribing the
+          # runner makes timing-sensitive tests (db-backup, upstream-timeout, ...) flake,
+          # which must not happen on a blocking gate. DATA_DIR isolation keeps the parallel
+          # run race-free regardless of concurrency.
+          if echo "$SEL" | grep -q "__RUN_ALL__"; then
+            echo "Fail-safe: running FULL unit suite (CI concurrency)"; npm run test:unit:ci; exit $?
+          fi
+          echo "Running impacted tests:"; echo "$SEL"
+          mapfile -t FILES <<< "$SEL"
+          node --import tsx --import ./open-sse/utils/setupPolyfill.ts --import ./tests/_setup/isolateDataDir.ts --test --test-force-exit --test-concurrency=4 "${FILES[@]}"

.github/workflows/scorecard.yml

@@ -0,0 +1,40 @@
+name: OpenSSF Scorecard
+on:
+  branch_protection_rule:
+  schedule:
+    - cron: "27 7 * * 1"
+  push:
+    branches: ["main"]
+
+permissions: read-all
+
+jobs:
+  analysis:
+    name: Scorecard analysis
+    runs-on: ubuntu-latest
+    permissions:
+      # security-events: write removed — Scorecard findings are advisory and no longer
+      # uploaded to the code-scanning Security tab (they are supply-chain/posture scores,
+      # not code vulnerabilities, and drowned out real CodeQL alerts). The run still
+      # produces the OpenSSF badge (publish_results) and a downloadable SARIF artifact.
+      id-token: write
+      contents: read
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+
+      - name: Run analysis
+        uses: ossf/scorecard-action@v2.4.3
+        with:
+          results_file: results.sarif
+          results_format: sarif
+          publish_results: true
+
+      - name: Upload artifact
+        uses: actions/upload-artifact@v7
+        with:
+          name: SARIF file
+          path: results.sarif
+          retention-days: 5

.github/workflows/semgrep.yml

@@ -0,0 +1,29 @@
+name: semgrep
+on:
+  pull_request:
+    branches: ["main", "release/**"]
+  push:
+    branches: ["main"]
+permissions:
+  contents: read
+jobs:
+  semgrep:
+    runs-on: ubuntu-latest
+    container:
+      image: semgrep/semgrep
+    steps:
+      - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        with:
+          persist-credentials: false
+      - name: Run semgrep (advisory)
+        continue-on-error: true
+        run: |
+          semgrep scan --config p/owasp-top-ten --config p/secrets \
+            --sarif --output semgrep.sarif --metrics off || true
+          python -c "import json; d=json.load(open('semgrep.sarif')); print('semgrepFindings=%d' % len(d['runs'][0]['results']))" || echo "semgrepFindings=SKIP"
+      - uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
+        if: always()
+        with:
+          name: semgrep-sarif
+          path: semgrep.sarif
+          retention-days: 14

.github/workflows/wiki-sync.yml

@@ -0,0 +1,69 @@
+name: Wiki Sync
+
+# Keeps the GitHub wiki in sync with docs/ on every release that lands on main.
+# The wiki has no native generator and historically drifts (it sat at "212+ providers /
+# 14 strategies / 37 MCP tools" while code was at 226 / 15 / 87, and new docs like
+# SUPPLY_CHAIN never appeared). This runs scripts/docs/sync-wiki.mjs, which:
+#   - ADDS any docs/ page missing from the wiki (curated; internal reports excluded),
+#   - syncs the four cover-page counts on Home.md.
+# It does NOT overwrite existing wiki pages by default: several docs sources still carry
+# stale counts (e.g. ARCHITECTURE.md says "177 providers" while the wiki cover is 226),
+# so blind overwrite would regress the wiki. Full content parity (--update-existing) is
+# gated on regenerating those sources — see docs/ops/DOCUMENTATION_AUDIT_REPORT.md.
+
+on:
+  push:
+    branches: [main]
+    paths:
+      - "docs/**"
+      - "README.md"
+      - "AGENTS.md"
+      - "src/shared/constants/routingStrategies.ts"
+      - "config/i18n.json"
+      - "open-sse/mcp-server/server.ts"
+      - "scripts/docs/sync-wiki.mjs"
+  workflow_dispatch:
+
+permissions:
+  contents: write
+
+concurrency:
+  group: wiki-sync
+  cancel-in-progress: false
+
+jobs:
+  sync-wiki:
+    name: Sync wiki with docs
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout repo
+        uses: actions/checkout@v7
+
+      - name: Setup Node
+        uses: actions/setup-node@v6
+        with:
+          node-version: "24"
+
+      - name: Clone wiki
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          REPO: ${{ github.repository }}
+        run: |
+          git clone "https://x-access-token:${GH_TOKEN}@github.com/${REPO}.wiki.git" wiki
+
+      - name: Sync wiki (add missing pages + cover counts)
+        run: node scripts/docs/sync-wiki.mjs --wiki-dir wiki
+
+      - name: Commit & push if changed
+        run: |
+          cd wiki
+          if [ -n "$(git status --porcelain)" ]; then
+            git config user.name "github-actions[bot]"
+            git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
+            git add -A
+            git commit -m "docs(wiki): auto-sync pages + cover counts with docs"
+            git push
+            echo "Wiki updated."
+          else
+            echo "Wiki already in sync — nothing to push."
+          fi

.gitignore

@@ -1,30 +1,67 @@
 # See https://help.github.com/articles/ignoring-files/ for more about ignoring files.
 
+# project-specific directories
+.omnivscodeagent/
+omnirouteCloud/
+omnirouteSite/
+_cache/
+_ideia/
+_mono_repo/
+_references/
+_tasks/
+.agents/**
+.claude/**
+.gemini/**
+.config/**
+.data/**
+.logs/**
+.tests/**
+.coverage/**
+coverage/
+.dist/**
+.next/**
+.build/**
+.out/**
+
+# Stryker mutation testing — ephemeral sandbox + generated reports (never commit)
+.stryker-tmp/
+reports/mutation/
+stryker-output-*.json
+
+
+# Memory Bank and Cursor rules (local-only AI agent context)
+memory-bank/
+.cursor/rules/core.mdc
+.cursor/rules/memory-bank.mdc
+
+# Claude Code local state — runtime files only; shared commands at .claude/commands/ are tracked
+.claude/scheduled_tasks.lock
+.claude/scheduled_tasks/
+.claude/sessions/
+.claude/state.json
+.claude/settings.local.json
+
+# Root-level underscore-prefixed directories (private/draft — never commit)
+/_*/
+
+# Draft features documentation (internal only)
+docs/new-features/
+
 # dependencies
+node_modules/
+# Also ignore a root node_modules SYMLINK (worktree setups symlink it from the main
+# checkout). The trailing-slash pattern above only matches a directory, so without this
+# a symlink named node_modules could be staged by `git add -A` and committed.
 /node_modules
-/.pnp
-.pnp.*
-.yarn/*
-!.yarn/patches
-!.yarn/plugins
-!.yarn/releases
-!.yarn/versions
-
-# testing
-/coverage
-
-# next.js
-/.next/
-/out/
-
-# production
-/build
-/app
-cloud/*
-
-# misc
+*.map
 .DS_Store
-*.pem
+
+# Obsidian sync plugin — committed for community distribution
+!obsidian-plugin/
+obsidian-plugin/node_modules/
+
+# Serena AI assistant config (local-only tool, not project code)
+.serena/
 
 # debug
 npm-debug.log*
@@ -35,6 +72,9 @@ yarn-error.log*
 # env files (can opt-in for committing if needed)
 .env*
 !.env.example
+# Provider API keys (never commit)
+*.api-key
+.nvidia-api-key
 
 # vercel
 .vercel
@@ -45,35 +85,15 @@ next-env.d.ts
 
 # data and logs
 data/
+.data/
 logs/*
+test_output.log
 
 # analysis directories (generated, not tracked)
 .analysis/
 antigravity-manager-analysis/
-
-# docs (allow specific tracked files)
-docs/*
-!docs/ARCHITECTURE.md
-!docs/CODEBASE_DOCUMENTATION.md
-!docs/CONTRIBUTING.md
-!docs/USER_GUIDE.md
-!docs/API_REFERENCE.md
-!docs/TROUBLESHOOTING.md
-!docs/EXECUTION_CONTEXT_PROVIDER_SYNC.md
-!docs/TASK_NEBIUS_BACKEND_ENABLEMENT.md
-!docs/frontend-backend-provider-gap-report.md
-!docs/openapi.yaml
-!docs/PLANO-IMPLANTACAO.md
-!docs/TASKS.md
-!docs/FASE-*.md
-!docs/adr/
-!docs/cli-tools/
-!docs/planning/
-!docs/improvement-plans/
-!docs/api/
-!docs/VM_DEPLOYMENT_GUIDE.md
-!docs/FEATURES.md
-!docs/screenshots/
+.sisyphus/
+.plans/
 
 # open-sse tests
 open-sse/test/*
@@ -82,16 +102,128 @@ open-sse/test/*
 .github/instructions/codacy.instructions.md
 
 # Playwright
+.playwright-mcp/
 test-results/
 playwright-report/
 blob-report/
 cloud/
-omnirouteCloud/
-omnirouteSite/
+.tmp/
 
 # Security Analysis (standalone project with own git)
 security-analysis/
 
 # Deploy workflow (contains sensitive VPS credentials)
-.agent/workflows/deploy.md
 clipr/
+app.log
+*.tgz
+.gh-discussions.json
+deploy.sh
+docker-compose.minimal.yml
+
+
+# Backup directories
+app.__qa_backup/
+.app-build-backup-*/
+backup/
+
+# Build intermediates (.build/) and shippable standalone (dist/).
+# These are fully reproducible from source; never committed.
+# Layer 1: Next.js now writes to .build/next (was .next); assembled bundle → dist/
+# (Previously /app/ was the standalone output; renamed to /dist/ in Layer 1.)
+/.build/
+/dist/
+/.next/
+
+# Electron
+electron/dist-electron/
+electron/node_modules/
+icon.iconset/
+
+# VS Code Extension (independent Git repo)
+vscode-extension/
+
+# SQLite residual files
+*.sqlite-shm
+*.sqlite-wal
+*.sqlite-journal
+
+# IDEA
+.idea/
+
+# Local OpenCode agent config
+.config/
+
+# Empty/dangling files
+typescript
+
+# Gemini Antigravity agent data
+.gemini/
+
+# Superpowers plans/specs (internal tooling, not project code)
+docs/superpowers/
+
+# TIA test-impact map — generated at runtime in CI (build-test-impact-map.mjs), never committed (~21MB)
+config/quality/test-impact-map.json
+
+# GitNexus local index
+.gitnexus
+.worktrees
+bin/omniroute.mjs
+
+# Consistent with .dockerignore / .npmignore
+.omc/
+audit-report.json
+bun.lock
+
+# Private environment variables for .http-client
+http-client.private.env.json
+
+# Note: _ideia/ (feature-triage drafts) is fully covered by the /_*/ rule above
+# and kept as a separate local-only git repo. Never committed to OmniRoute.
+
+# i18n audit artifact (generated by scripts/i18n/audit-dashboard-pages.mjs)
+scripts/i18n/_audit.json
+scripts/i18n/_pending-keys.json
+
+# Private workflow / skill / command implementations
+# These contain proprietary multi-phase logic and should not be committed
+.agents/workflows/implement-features-ag.md
+.agents/workflows/port-upstream-features-ag.md
+.agents/workflows/port-upstream-issues-ag.md
+.agents/skills/implement-features/
+.claude/commands/implement-features-cc.md
+.claude/commands/port-upstream-features-cc.md
+.claude/commands/port-upstream-issues-cc.md
+.claude/worktrees/
+.codegraph/
+
+# Fumadocs generated source
+.source/
+
+# AI agent local settings and configs
+.agents/
+.antigravitycli/
+.claude/
+
+# PR Reviews and local feedback files
+pr_reviews*.json
+
+#hidden local data directories (never commit)
+.local-data/
+.data-dev/
+/.junie/
+
+# internal setup prompts with personal credentials — never commit
+CODEX-SETUP-PROMPT.md
+# Quality ratchet — métricas efêmeras (baseline commitado em config/quality/; métricas não)
+config/quality/quality-metrics.json
+
+# Runtime logs (diretório local, nunca versionado)
+/logs/
+-home-diegosouzapw-dev-automações-bots-yt-downloader-20260504 .txt
+-home-diegosouzapw-dev-automações-bots-yt-downloader-20260410 .txt
+docs/prompts/AGENT-OWNERSHIP-PROTOCOL.omniroute.md
+docs/prompts/AGENT-OWNERSHIP-PROTOCOL.md
+docs/prompts/AGENT-OWNERSHIP-PROTOCOL.omniroute-mim.md
+docs/prompts/AGENT-OWNERSHIP-PROTOCOL.omniroute-mid.md
+omniroute.md

.gitleaks.toml

@@ -0,0 +1,76 @@
+# .gitleaks.toml — Configuração do gitleaks para OmniRoute
+# Task 7.18 — PLANO-QUALITY-GATES-FASE7.md
+#
+# Estende as regras padrão do gitleaks com allowlists específicas do projeto.
+#
+# INSTRUÇÕES para allowlists:
+#   Findings legítimos (fixtures de teste, creds OAuth públicas já cobertas pelo
+#   check-public-creds.mjs, valores de exemplo em docs) devem ser registrados abaixo
+#   em [[allowlist]] com um comentário explicativo obrigatório.
+#
+#   NÃO adicione uma entrada de allowlist sem justificativa. Cada entrada é revisada
+#   a cada release (stale-enforcement). Regra: se o finding é um valor real que o
+#   sistema usa em produção, é um verdadeiro positivo — não allowliste, corrija.
+#
+# Referência: docs/security/PUBLIC_CREDS.md (credenciais OAuth públicas conhecidas)
+#             CLAUDE.md Hard Rule #11 (resolvePublicCred obrigatório)
+
+# Herdar TODAS as regras padrão do gitleaks. ATENÇÃO: um config customizado
+# SEM [extend].useDefault = true (e sem [[rules]] próprias) resulta em ZERO
+# regras — o gitleaks SUBSTITUI o ruleset padrão pelo arquivo, não o estende
+# automaticamente. Sem esta seção, `gitleaks --config .gitleaks.toml` nunca
+# detecta nada (todo finding vira 0), tornando o gate inerte. Com useDefault,
+# a allowlist abaixo é aplicada POR CIMA das ~170 regras padrão.
+[extend]
+  useDefault = true
+
+# Para desabilitar uma regra específica, usar:
+# [[rules]]
+#   id = "rule-id"
+#   [rules.allowlist]
+#     description = "..."
+
+# ---------------------------------------------------------------------------
+# Allowlist global do projeto
+# Entradas aqui são ignoradas em TODAS as varreduras.
+# ---------------------------------------------------------------------------
+[allowlist]
+  description = "OmniRoute project-level allowlist — fixtures, test vectors, public OAuth creds"
+
+  # Paths a ignorar completamente (node_modules, builds, etc.)
+  paths = [
+    '''node_modules''',
+    '''\.next''',
+    '''dist''',
+    '''\.git''',
+    '''coverage''',
+    '''\.nyc_output''',
+  ]
+
+  # Commits específicos a ignorar (ex: commit que introduziu fixtures de teste)
+  # commits = []
+
+  # Regexes de stopwords — linhas que contêm estes padrões são ignoradas.
+  # Usar apenas para falsos positivos comprovados com justificativa abaixo.
+  # stopwords = []
+
+  # Regexes de targets (paths de arquivos) que podem ser allowlistados por regra.
+  # Ver [[allowlist]] por-regra abaixo para granularidade.
+
+# ---------------------------------------------------------------------------
+# Allowlist por-regra (adicionar conforme necessário durante o stale review)
+# ---------------------------------------------------------------------------
+#
+# Exemplo (REMOVER / SUBSTITUIR por entradas reais quando necessário):
+#
+# [[rules]]
+#   # Allowlistar fixtures de teste que contêm tokens OAuth de exemplo/inválidos
+#   # Adicionado: 2026-06-13 | Revisar em: v3.9.0 | Justificativa: valores não-reais de teste
+#   id = "github-fine-grained-pat"
+#   [rules.allowlist]
+#     description = "Test fixture PATs — valores sintéticos, não funcionais"
+#     paths = [
+#       '''tests/fixtures/''',
+#       '''tests/unit/''',
+#     ]
+#

.husky/pre-commit

@@ -1 +1,13 @@
+#!/usr/bin/env sh
+if ! command -v npx >/dev/null 2>&1; then
+  echo "⚠️  npx not found in PATH — skipping pre-commit hooks"
+  echo "   Run 'npm run lint && npm run check:any-budget:t11' manually before pushing."
+  exit 0
+fi
+
+# Cheap, deterministic local gates (re-enabled). Slower checks (i18n drift,
+# openapi coverage/security-tiers, env-doc sync) run in CI to keep commits fast.
 npx lint-staged
+node scripts/check/check-docs-sync.mjs
+npm run check:any-budget:t11
+node scripts/check/check-tracked-artifacts.mjs

.husky/pre-push

@@ -0,0 +1,12 @@
+#!/usr/bin/env sh
+# .husky/pre-push — fast deterministic gates (<10s total)
+# Intentionally excludes test:unit (slow; covered by CI pre-push remote run).
+# Activated: 2026-06-13 (6A.12 — replaced commented-out test:unit stub)
+
+if ! command -v npm >/dev/null 2>&1; then
+  echo "⚠️  npm not found in PATH — skipping pre-push hooks"
+  echo "   Run 'npm run check:any-budget:t11 && npm run check:tracked-artifacts' manually before pushing."
+  exit 0
+fi
+
+npm run check:any-budget:t11 && npm run check:tracked-artifacts

.markdownlint.json

@@ -0,0 +1,14 @@
+{
+  "_comment": "Advisory markdown lint for docs/ + root *.md. Rules that conflict with the existing doc style (heavy inline HTML, long lines, centered headings) are disabled so the gate stays signal-not-noise. Run: npm run lint:md",
+  "default": true,
+  "MD013": false,
+  "MD033": false,
+  "MD041": false,
+  "MD024": { "siblings_only": true },
+  "MD026": false,
+  "MD036": false,
+  "MD040": false,
+  "MD029": false,
+  "MD007": { "indent": 2 },
+  "MD046": false
+}

.mcp.json.example

@@ -0,0 +1,17 @@
+{
+  "$comment_purpose": "OPT-IN agent-lsp / LSP-in-the-loop (Quality Gates Fase 7 Task 15). Copy this file to `.mcp.json` to enable. It exposes a TypeScript language server to coding agents (Claude Code, etc.) so they get diagnostics / hover / go-to-definition / blast-radius BEFORE writing code — turning 'invented symbol' review-catches into impossible-at-edit-time. Pairs with `npm run typecheck:core` as a compile-before-claim check.",
+  "$comment_safety": "Shipped as `.example` (NOT `.mcp.json`) on purpose so it never auto-loads an unvetted server into everyone's session. Pick an MCP<->LSP bridge you trust and have verified locally, then drop in its package + args below. A broken MCP entry only logs a connection error; it does not break agent sessions. The underlying language server is `typescript-language-server` (npm, mature) — install via `npm i -g typescript-language-server typescript` or rely on npx.",
+  "mcpServers": {
+    "typescript-lsp": {
+      "command": "npx",
+      "args": [
+        "-y",
+        "<your-mcp-lsp-bridge>",
+        "--lsp",
+        "typescript-language-server",
+        "--stdio"
+      ],
+      "$note": "Replace <your-mcp-lsp-bridge> with the concrete MCP<->LSP adapter you chose. It must speak MCP on stdio and proxy to `typescript-language-server --stdio`. Scope it to this repo's tsconfig (open-sse/tsconfig.json / tsconfig.json) for accurate diagnostics."
+    }
+  }
+}

.node-version

@@ -0,0 +1 @@
+24

.npmignore

@@ -3,7 +3,22 @@ data/
 **/data/
 **/db.json
 
+# VS Code extension test runtime (large binary, not needed in npm package)
+app/vscode-extension/
+**/data/
+**/db.json
+
 # Source code (pre-built app/ is published instead)
+#
+# NOTE (#3578 / #3821-review): package.json "files" is the source of truth for what
+# ships. It now allowlists the backend source closure the MCP server needs at runtime
+# (open-sse/, src/lib, src/server, ...) and OVERRIDES the broad src/ + open-sse/ excludes
+# below — npm honors files[] over .npmignore for inclusion. These lines are kept only as
+# intent/back-stop: if files[] is ever trimmed back to specific paths, they must NOT be
+# allowed to re-hide the MCP closure (that would silently reintroduce the --mcp
+# ERR_MODULE_NOT_FOUND #3578 fixed). The closure gate in
+# tests/unit/mcp-published-files-closure-3578.test.ts asserts the real `npm pack` output
+# in both directions (closure present + zero test files), catching such a regression.
 src/
 open-sse/
 docs/
@@ -13,6 +28,17 @@ images/
 logs/
 scripts/
 
+# Co-located tests must never ship even when their parent dir is allowlisted by files[].
+# (Primary guard is the "!**/*.test.*" negations in package.json files[]; this is defense
+# in depth for any nested dir the allowlist pulls in.)
+**/__tests__/
+**/*.test.ts
+**/*.test.tsx
+**/*.test.js
+**/*.test.mjs
+**/*.spec.ts
+**/*.spec.tsx
+
 # Config/dev files
 *.md
 !README.md
@@ -21,14 +47,21 @@ scripts/
 .github/
 .husky/
 .vscode/
+.agents/
 .env*
+app/.env
+app/.env*
 eslint.config.mjs
 prettier.config.mjs
 postcss.config.mjs
 next.config.mjs
 tsconfig.json
+tsconfig.typecheck-core.json
+tsconfig.typecheck-noimplicit-core.json
 playwright.config.ts
+vitest.config.ts
 next-env.d.ts
+llm.txt
 
 # Docker
 docker-compose*.yml
@@ -36,9 +69,56 @@ Dockerfile
 .dockerignore
 
 # Misc
-restart.sh
 AGENTS.md
+bun.lock
 
 # Build artifacts (pre-built goes inside app/)
-.next/
-node_modules/
+/.next/
+/node_modules/
+
+# Ignore large binary files and other build directories
+*.tgz
+*.AppImage
+*.deb
+*.rpm
+electron/
+app/electron/
+app/vscode-extension/
+
+# Subprojects
+clipr/
+omnirouteCloud/
+omnirouteSite/
+vscode-extension/
+
+# Root-level underscore-prefixed directories (private/draft — never publish)
+/_*/
+app/_*/
+app/coverage/
+app/logs/
+app/tests/
+
+# Consistent with .gitignore and .dockerignore
+.DS_Store
+.idea/
+.config/
+.data/
+.omnivscodeagent/
+.omc/
+*.sqlite-*
+*.tsbuildinfo
+security-analysis/
+.analysis/
+antigravity-manager-analysis/
+.sisyphus/
+.plans/
+app.__qa_backup/
+.app-build-backup-*/
+.gitnexus
+.worktrees
+.next-playwright/
+test-results/
+playwright-report/
+blob-report/
+coverage/
+@omniroute/

.npmrc

@@ -0,0 +1,4 @@
+# @lobehub/icons declares UI peers that are not needed by our deep icon imports.
+# Keeping peer auto-install disabled prevents npm from pulling @lobehub/ui/mermaid
+# back into the tree and reopening npm audit findings for unused packages.
+legacy-peer-deps=true

.nvmrc

@@ -0,0 +1 @@
+24

.size-limit.json

@@ -0,0 +1,22 @@
+[
+  {
+    "name": "CLI entry (omniroute.mjs)",
+    "path": "bin/omniroute.mjs",
+    "limit": "15 KB"
+  },
+  {
+    "name": "MCP server entry (mcp-server.mjs)",
+    "path": "bin/mcp-server.mjs",
+    "limit": "5 KB"
+  },
+  {
+    "name": "Node runtime support (nodeRuntimeSupport.mjs)",
+    "path": "bin/nodeRuntimeSupport.mjs",
+    "limit": "8 KB"
+  },
+  {
+    "name": "Reset password entry (reset-password.mjs)",
+    "path": "bin/reset-password.mjs",
+    "limit": "6 KB"
+  }
+]

.source/dynamic.ts

@@ -0,0 +1,8 @@
+// @ts-nocheck
+import { dynamic } from 'fumadocs-mdx/runtime/dynamic';
+import * as Config from '../source.config';
+
+const create = await dynamic<typeof Config, import("fumadocs-mdx/runtime/types").InternalTypeConfig & {
+  DocData: {
+  }
+}>(Config, {"configPath":"source.config.ts","environment":"next","outDir":".source"}, {"doc":{"passthroughs":["extractedReferences"]}});

.source/source.config.mjs

@@ -0,0 +1,22 @@
+// source.config.ts
+import { defineDocs, defineConfig } from "fumadocs-mdx/config";
+var docs = defineDocs({
+  dir: "docs",
+  docs: {
+    files: [
+      "./architecture/**/*.md",
+      "./guides/**/*.md",
+      "./reference/**/*.md",
+      "./frameworks/**/*.md",
+      "./routing/**/*.md",
+      "./security/**/*.md",
+      "./compression/**/*.md",
+      "./ops/**/*.md"
+    ]
+  }
+});
+var source_config_default = defineConfig();
+export {
+  source_config_default as default,
+  docs
+};

.vale.ini

@@ -0,0 +1,15 @@
+# Advisory prose lint for OmniRoute docs (Vale — https://vale.sh).
+# Warning-first: surfaces vague language, passive voice, and terminology drift without
+# blocking CI. The Microsoft style is pulled by the CI step (vale sync / vale-action).
+# Project terms live in the OmniRoute vocabulary so they never read as misspellings.
+StylesPath = .vale/styles
+MinAlertLevel = warning
+Packages = Microsoft
+Vocab = OmniRoute
+
+[*.md]
+BasedOnStyles = Vale, Microsoft
+
+# Code-heavy reference docs and auto-generated catalogs are noisy under prose rules.
+[docs/reference/PROVIDER_REFERENCE.md]
+BasedOnStyles = Vale

.vale/styles/Vocab/OmniRoute/accept.txt

@@ -0,0 +1,50 @@
+OmniRoute
+combo
+combos
+[Cc]ombo
+[Aa]uto-[Cc]ombo
+MCP
+A2A
+ACP
+RTK
+Caveman
+Troglodita
+Qdrant
+FTS5
+SQLite
+LowDB
+OAuth
+PKCE
+SSE
+JSON-RPC
+LKGP
+P2C
+Antigravity
+Codex
+Kiro
+Qoder
+Pollinations
+LongCat
+Cerebras
+DeepSeek
+Groq
+Cline
+OpenCode
+Termux
+Electron
+Fumadocs
+Notion
+Obsidian
+WebDAV
+IPv4
+IPv6
+SOCKS5
+loopback
+backoff
+changelog
+i18n
+locale
+locales
+roadmap
+[Ww]ildcard
+upstream

.vscode/settings.json

@@ -1,4 +1,5 @@
 {
+  "workbench.sideBar.location": "left",
   "css.lint.unknownAtRules": "ignore",
   "sonarlint.rules": {
     "css:S4662": {
@@ -16,5 +17,118 @@
     "javascript:S3776": {
       "level": "off"
     }
+  },
+  "git.ignoreLimitWarning": true,
+
+  // ─── Git: não adicionar os ~44 repos aninhados (9 worktrees + _references/* +
+  // _mono_repo/* + _ideia/_tasks/.agents) ao Source Control. Era a causa do
+  // "validando muito": um git status + watcher por repo. Só o repo raiz fica. ───
+  "git.autoRepositoryDetection": false,
+  "git.repositoryScanMaxDepth": 0,
+  "git.detectSubmodules": false,
+  "git.autofetch": false,
+  "git.autorefresh": true,
+
+  // ─── Performance: não seguir o symlink de node_modules criado pelas worktrees ───
+  // As worktrees em .worktrees/ apontam node_modules para o checkout principal;
+  // sem isto a busca atravessa o symlink mesmo com node_modules excluído.
+  "search.followSymlinks": false,
+
+  // ─── TypeScript server (monorepo grande: ~6,6k arquivos rastreados, Next.js 16) ───
+  "typescript.tsserver.maxTsServerMemory": 4096,
+  "typescript.disableAutomaticTypeAcquisition": true,
+  "typescript.tsc.autoDetect": "off",
+  "npm.autoDetect": "off",
+  // O tsserver tem o próprio watcher (independente de files.watcherExclude).
+  // excludeDirectories evita que ele vigie as árvores pesadas.
+  "typescript.tsserver.watchOptions": {
+    "excludeDirectories": [
+      "**/node_modules",
+      "**/.next",
+      "**/.build",
+      "**/dist",
+      "**/coverage",
+      "**/.worktrees"
+    ]
+  },
+
+  // Para esconder os diretórios gerados da árvore do Explorer, descomente:
+  // "files.exclude": {
+  //   "**/.worktrees": true,
+  //   "**/coverage": true,
+  //   "**/dist": true,
+  //   "**/.build": true,
+  //   "**/.next": true,
+  //   "**/_references": true,
+  //   "**/_mono_repo": true,
+  //   "**/_tasks": true,
+  //   "**/omniroute-*.tgz": true
+  // },
+  "files.watcherExclude": {
+    "**/.git/objects/**": true,
+    "**/.git/subtree-cache/**": true,
+    "**/node_modules/**": true,
+    "**/.next/**": true,
+    "**/.build/**": true,
+    "**/dist/**": true,
+    "**/build/**": true,
+    "**/out/**": true,
+    "**/coverage/**": true,
+    "**/.coverage/**": true,
+    "**/.nyc_output/**": true,
+    "**/.cache/**": true,
+    "**/.turbo/**": true,
+    "**/.swc/**": true,
+    "**/.stryker-tmp/**": true,
+    "**/stryker-output-*/**": true,
+    "**/playwright-report/**": true,
+    "**/test-results/**": true,
+    "**/.worktrees/**": true,
+    "**/.claude/worktrees/**": true,
+    "**/electron/**": true,
+    "**/_references/**": true,
+    "**/_mono_repo/**": true,
+    "**/_tasks/**": true,
+    "**/logs/**": true,
+    "**/*.tgz": true,
+    "**/*.tsbuildinfo": true,
+    "**/OmniRoute-*/**": true,
+    "**/*-merge-*/**": true,
+    "**/*-worktree-*/**": true,
+    "**/*-issues-*/**": true,
+    "**/*-reorg*/**": true
+  },
+  "search.exclude": {
+    "**/node_modules": true,
+    "**/.next": true,
+    "**/.build": true,
+    "**/dist": true,
+    "**/build": true,
+    "**/out": true,
+    "**/coverage": true,
+    "**/.coverage": true,
+    "**/.nyc_output": true,
+    "**/.cache": true,
+    "**/.turbo": true,
+    "**/.swc": true,
+    "**/.stryker-tmp": true,
+    "**/stryker-output-*": true,
+    "**/playwright-report": true,
+    "**/test-results": true,
+    "**/.worktrees": true,
+    "**/.claude/worktrees": true,
+    "**/electron": true,
+    "**/_references": true,
+    "**/_mono_repo": true,
+    "**/_tasks": true,
+    "**/logs": true,
+    "**/*.tgz": true,
+    "**/*.tsbuildinfo": true,
+    "**/package-lock.json": true,
+    "**/OmniRoute-*": true,
+    "**/*-merge-*": true,
+    "**/*-worktree-*": true,
+    "**/*-issues-*": true,
+    "**/*-reorg*": true
   }
 }

.zizmor.yml

@@ -0,0 +1,104 @@
+# .zizmor.yml — zizmor security audit configuration
+# https://github.com/woodruffw/zizmor
+#
+# zizmor audits GitHub Actions workflows for security anti-patterns:
+#   • Unpinned third-party actions (supply-chain risk)
+#   • Script injection via ${{ github.* }} in run: steps
+#   • pull_request_target misuse (allows untrusted code to reach secrets)
+#   • Excessive permissions / overly broad GITHUB_TOKEN scopes
+#   • Cache poisoning via actions/cache with unguarded keys
+#   • … 20+ additional audits
+#
+# MOTIVATION (2026-03 incident): the trivy-action/LiteLLM supply-chain attack
+# exploited exactly the kind of pull_request_target misconfiguration that zizmor
+# detects statically. OmniRoute's npm/Docker/Electron publish workflows are high-
+# value targets that warrant proactive audit.
+#
+# BASELINE STRATEGY: this file starts with an EMPTY ignores list. As zizmor is
+# first run against the existing workflows, findings will be reviewed:
+#   • True positives → fix the workflow (preferred)
+#   • Accepted risk  → add an entry below with a justification comment
+# This enforces the same stale-enforcement discipline as other allowlists in
+# this project: every ignore requires human sign-off and is subject to review
+# at each release cycle.
+#
+# RATCHET: zizmorFindings metric in quality-baseline.json (direction: down)
+# starts advisory; current baseline is frozen at first green run; new findings
+# must be remediated or explicitly ignored here.
+
+# ── Global settings ──────────────────────────────────────────────────────────
+# Uncomment to pin to a specific minimum severity level.
+# min-severity: low       # low | medium | high | critical (default: low = all)
+
+# ── Per-rule ignores ─────────────────────────────────────────────────────────
+# zizmor ≥1.0 replaced the old top-level `ignores:` list with a `rules:` map
+# keyed by audit id. Per-rule config lives under `rules.<audit-id>.ignore`,
+# where each entry is `filename` or `filename:line` (line optional, column
+# further-optional). See: https://docs.zizmor.sh/configuration/
+#
+# Format:
+#   rules:
+#     <zizmor-audit-id>:          # e.g. "unpinned-uses", "template-injection"
+#       ignore:
+#         - <workflow-filename>            # ignore this audit across the file
+#         - <workflow-filename>:<line>     # or scope to a specific line
+#
+# Example (do not uncomment without real justification):
+#
+#   rules:
+#     unpinned-uses:
+#       # actions/checkout@v6 is pinned at the major-version tag intentionally:
+#       # GitHub-managed first-party action; SHA pinning buys little against a
+#       # GitHub-side compromise and adds significant maintenance burden.
+#       ignore:
+#         - ci.yml
+
+# Every entry below is an explicit, justified ignore (same stale-review
+# discipline as the other allowlists in this project). New ignores require a
+# per-rule entry with a justification comment.
+rules:
+  dangerous-triggers:
+    # deploy-vps.yml uses `on: workflow_run` (after "Publish to Docker Hub").
+    # zizmor flags workflow_run as "almost always used insecurely", but this one
+    # is guarded and not exploitable:
+    #   • The deploy job is gated on `github.event.workflow_run.conclusion ==
+    #     'success'` (deploy-vps.yml job `if:`, L15), so it only runs after a
+    #     successful, trusted upstream run — never on a forked/PR-triggered
+    #     failure.
+    #   • It does NOT checkout or execute untrusted code: the deploy runs over SSH
+    #     using repository secrets; there is no `actions/checkout` of an attacker
+    #     ref in the privileged context.
+    # Accepted risk, re-review at the next release cycle (added 2026-06-15).
+    ignore:
+      - deploy-vps.yml
+
+  cache-poisoning:
+    # zizmor's cache-poisoning audit flags `actions/setup-node` (cache: npm) and
+    # `actions/cache` steps that populate a cache which a later artifact-publishing
+    # job could consume — the attack is: a fork PR seeds a poisoned cache that a
+    # trusted publish run then restores and ships. None of the four entries below
+    # are exploitable in that way:
+    #
+    #   • electron-release.yml / npm-publish.yml — PUBLISH workflows that trigger
+    #     ONLY on trusted events: `push: tags v*` + `workflow_dispatch` (electron)
+    #     and `release: [released]` + `workflow_dispatch` + `workflow_call`
+    #     (npm). They NEVER run on a `pull_request` from a fork, so an untrusted
+    #     ref can never populate the npm/node_modules cache they restore. The
+    #     cache key is `hashFiles('package-lock.json')` / `runner.os-node-…`, a
+    #     first-party, lockfile-pinned key on trusted events.
+    #
+    #   • opencode-plugin-ci.yml / opencode-provider-ci.yml — CI-only workflows
+    #     (lint/test of the two opencode packages). They DO run on `pull_request`,
+    #     but (a) they publish NO artifacts — there is no downstream publish job
+    #     that restores their cache, and (b) GitHub Actions cache is branch-scoped:
+    #     a fork-PR run writes to a PR-isolated cache that base-branch / release
+    #     runs cannot read. The `cache: npm` here only speeds up `npm ci` within
+    #     the same ephemeral CI run.
+    #
+    # Accepted risk (first-party caching on trusted/isolated events), re-review at
+    # the next release cycle (added 2026-06-16).
+    ignore:
+      - electron-release.yml
+      - npm-publish.yml
+      - opencode-plugin-ci.yml
+      - opencode-provider-ci.yml

@omniroute/opencode-plugin/.gitignore

@@ -0,0 +1,4 @@
+node_modules
+dist
+*.log
+.DS_Store

@omniroute/opencode-plugin/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 OmniRoute contributors
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

@omniroute/opencode-plugin/README.md

@@ -0,0 +1,296 @@
+# @omniroute/opencode-plugin
+
+> **Recommended way to use OmniRoute with OpenCode.** Pulls a live model catalog from `/v1/models` (including `-low`/`-medium`/`-high`/`-thinking` variants as first-class IDs), aggregates combos via `/api/combos` using a least-common-denominator capability/limit join, sanitizes Gemini tool schemas in flight, and supports multiple side-by-side OmniRoute instances out of the box.
+
+## Why this and not `@omniroute/opencode-provider`?
+
+`@omniroute/opencode-provider` is the legacy config-generator package — it writes a frozen `provider.omniroute` block into `opencode.json` with a **hardcoded list of 8 models** ([`OMNIROUTE_DEFAULT_OPENCODE_MODELS`](https://github.com/diegosouzapw/OmniRoute/blob/main/%40omniroute/opencode-provider/src/index.ts#L48-L56)). It works on the CLI but in the **OpenCode Desktop / Web** builds (Tauri / Electron) the runtime re-runs the model picker and the static block surfaces only a few of those — and they drift behind the live OmniRoute catalog.
+
+This plugin solves that by:
+
+- Fetching `/v1/models` and `/api/combos` **at OpenCode startup, in Node.js** — no CORS, no WebView restrictions
+- Emitting the provider block **dynamically** in the plugin's `config`/`provider` hook — so `opencode.json` only needs the plugin entry, not a static `provider.omniroute`
+- Re-fetching on a configurable TTL (default 5 min), so new models / combo changes in the OmniRoute UI appear without restarting OpenCode
+- Computing `limit.context` for combos as `min(member.context_length)` from the live catalog (no more `null` values that cause 4K-token truncation)
+- **Auto-pickup of `interleaved` capability** for thinking models (merged via PR #3138)
+
+**If you only have the legacy `opencode-provider` block in your `opencode.json`, replace it with a single plugin entry.** No other config changes required — the same `auth.json` API key works.
+
+## Install
+
+The plugin ships **pre-built inside the `omniroute` npm package** since v3.8.23.
+If you have OmniRoute installed, the plugin is already on disk:
+
+```sh
+# 1. One command — copy the plugin into OpenCode and update opencode.json
+omniroute setup opencode --auth
+
+# 2. Follow the interactive prompt to enter your OmniRoute API key
+# 3. Restart OpenCode — /models lists the full live catalog
+```
+
+The `--auth` flag runs `opencode auth login --provider omniroute` automatically.
+Use `--base-url` to point at a non-default OmniRoute address:
+
+```sh
+omniroute setup opencode --base-url https://or.example.com --auth
+```
+
+### What it does
+
+1. Locates the bundled plugin inside the omniroute installation
+2. Copies `dist/` + `package.json` to `~/.config/opencode/plugins/omniroute/`
+3. Writes/updates `opencode.json` with the plugin entry (idempotent, replaces legacy entries)
+4. (With `--auth`) runs `opencode auth login` so the API key is stored
+
+Re-run any time to update the plugin or change the base URL. Older entries for
+`@omniroute/opencode-provider` or the legacy `opencode-omniroute-auth` package are
+automatically cleaned up.
+
+### Manual install (without omniroute CLI)
+
+If you cannot run `omniroute setup opencode` (local dev, CI, air-gapped), reference
+the built artifact directly:
+
+```sh
+cd @omniroute/opencode-plugin && npm run build && npm pack
+# then extract into ~/.config/opencode/plugins/omniroute-opencode-plugin/
+```
+
+And add the entry to `opencode.json` manually (see Quick Start below).
+
+Peer dep: `@opencode-ai/plugin` (managed by your OpenCode install).
+
+## Quick start (single instance, manual)
+
+```jsonc
+// opencode.json
+{
+  "$schema": "https://opencode.ai/config.json",
+  "plugin": [
+    [
+      "./plugins/omniroute-opencode-plugin/dist/index.js",
+      {
+        "providerId": "omniroute",
+        "baseURL": "https://or.example.com",
+      },
+    ],
+  ],
+}
+```
+
+```sh
+opencode auth login --provider omniroute
+# prompts for the OmniRoute API key, writes to ~/.local/share/opencode/auth.json
+```
+
+> ⚠ Use the `--provider` flag explicitly. `opencode auth login omniroute` is parsed as a positional `url` argument by current OC releases (≤1.15.5) and fails with `fetch() URL is invalid`. Tracked upstream.
+
+Restart OpenCode. `/models` lists the full live catalog. Variants (`-low`, `-medium`, `-high`, `-thinking`) and combos appear as first-class IDs — OmniRoute is the source of truth, no client-side synthesis.
+
+## Multi-instance (prod + preprod side-by-side)
+
+> ⚠ OC ≤1.15.5 dedupes plugin loads by absolute module path. Two `plugin:` entries pointing at the same `dist/index.js` collapse into one (last-listed options win). Workaround: install the plugin twice into separate directories so each entry resolves to a distinct module file. v0.2.x will introduce an `instances: [...]` shape that registers N providers from a single load.
+
+### Dual-install workaround (works today on OC ≤1.15.5)
+
+Pack the plugin once, extract it twice into named directories, then point each `plugin:` entry at its own copy:
+
+```sh
+# 1. Build + pack the plugin (run from the plugin worktree)
+cd /path/to/OmniRoute/@omniroute/opencode-plugin
+npm run build
+npm pack
+# produces omniroute-opencode-plugin-0.1.0.tgz
+
+# 2. Extract one copy per OmniRoute endpoint
+mkdir -p ~/.config/opencode/plugins/omniroute-opencode-plugin-prod
+mkdir -p ~/.config/opencode/plugins/omniroute-opencode-plugin-preprod
+tar -xzf omniroute-opencode-plugin-0.1.0.tgz -C ~/.config/opencode/plugins/omniroute-opencode-plugin-prod    --strip-components=1
+tar -xzf omniroute-opencode-plugin-0.1.0.tgz -C ~/.config/opencode/plugins/omniroute-opencode-plugin-preprod --strip-components=1
+```
+
+Then in `~/.config/opencode/opencode.json` reference each directory by absolute path:
+
+```jsonc
+{
+  "$schema": "https://opencode.ai/config.json",
+  "plugin": [
+    [
+      "./plugins/omniroute-opencode-plugin-prod/dist/index.js",
+      {
+        "providerId": "omniroute",
+        "displayName": "OmniRoute",
+        "baseURL": "https://or.example.com",
+      },
+    ],
+    [
+      "./plugins/omniroute-opencode-plugin-preprod/dist/index.js",
+      {
+        "providerId": "omniroute-preprod",
+        "displayName": "OmniRoute Preprod",
+        "baseURL": "https://or-preprod.example.com",
+      },
+    ],
+  ],
+}
+```
+
+Paths are relative to `~/.config/opencode/`. Each entry now resolves to a distinct module file, so OC loads them as two separate plugin instances. Authenticate each:
+
+```sh
+opencode auth login --provider omniroute
+opencode auth login --provider omniroute-preprod
+```
+
+Each entry gets its own provider id, its own model picker entry, its own slot in `auth.json`, and its own TTL cache. Closures are isolated per plugin instance — no cross-talk.
+
+### After publish (`@omniroute/opencode-plugin` npm)
+
+Once the package is published, the dual-install becomes two `npm install --prefix` commands instead of `tar -xzf`:
+
+```sh
+mkdir -p ~/.config/opencode/plugins/omniroute-opencode-plugin-prod
+mkdir -p ~/.config/opencode/plugins/omniroute-opencode-plugin-preprod
+npm install --prefix ~/.config/opencode/plugins/omniroute-opencode-plugin-prod    @omniroute/opencode-plugin
+npm install --prefix ~/.config/opencode/plugins/omniroute-opencode-plugin-preprod @omniroute/opencode-plugin
+```
+
+`opencode.json` paths become `./plugins/omniroute-opencode-plugin-prod/node_modules/@omniroute/opencode-plugin/dist/index.js` (and the preprod equivalent).
+
+## Features
+
+| Feature                                     | What it does                                                                                                                                                                                                                                                                                                                                                                                                | Hook                         |
+| ------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------- |
+| Dynamic `/v1/models`                        | Pulls live catalog (455+ entries on prod) on each refresh, TTL-cached                                                                                                                                                                                                                                                                                                                                       | `provider.models`            |
+| Variants pass-through                       | `-low`/`-medium`/`-high`/`-thinking` ship as first-class IDs from OmniRoute (no client synthesis)                                                                                                                                                                                                                                                                                                           | `provider.models`            |
+| Combo LCD aggregation                       | Combos appear with intersected capabilities + min context/output across members                                                                                                                                                                                                                                                                                                                             | `provider.models` + `config` |
+| `combo/<slug>` namespace + `Combo: ` prefix | Combos surface under `combo/claude-primary` (not the upstream UUID) and the picker shows `Combo: claude-primary` so they stand apart from raw provider/model pairs                                                                                                                                                                                                                                          | both hooks                   |
+| Nice names + cost                           | `/api/pricing/models` display names AND `/api/pricing` per-million-token cost overlaid onto the live catalog                                                                                                                                                                                                                                                                                                | both hooks                   |
+| Canonical-twin dedup + alias-fallback       | `/v1/models` exposes the same upstream model under both short alias (`cc/claude-opus-4-7`) and canonical name (`claude/claude-opus-4-7`); the plugin drops the canonical twin when an alias twin exists (no duplicate rows in the picker) and reverse-maps canonical → alias to pick up enrichment for short aliases (`dg/nova-3 → Deepgram - Nova 3`) that `/api/pricing/models` only indexes by canonical | both hooks                   |
+| Compression pipeline tags                   | Combo names get tagged with their compression pipeline (e.g. `Combo: claude-primary [rtk🟡 → caveman🟠]`) when `features.compressionMetadata: true`. Intensity tokens render as a traffic-light emoji: 🟢 lite/minimal · 🟡 standard · 🟠 aggressive/full · 🔴 ultra                                                                                                                                        | both hooks                   |
+| Provider-tag prefix                         | Prepend short upstream-provider label to enriched names (e.g. `Claude - Claude Opus 4.7` vs `Kiro - Claude Opus 4.7`, `GHM - GPT 5`) so same-id models routed via different upstream connections group visibly in the picker (default-on, opt-out via `features.providerTag: false`)                                                                                                                        | both hooks                   |
+| Usable-only filter                          | Filter to providers with at least one healthy connection in `/api/providers` (opt-in via `features.usableOnly`)                                                                                                                                                                                                                                                                                             | both hooks                   |
+| Disk-cache fallback                         | Last-known-good catalog persisted to disk; hydrates on a cold start when `/v1/models` is unreachable (default-on, opt-out via `features.diskCache: false`)                                                                                                                                                                                                                                                  | `config`                     |
+| Bearer injection + suffix-spoof guard       | Adds `Authorization` on baseURL-matched requests only                                                                                                                                                                                                                                                                                                                                                       | `auth.loader.fetch`          |
+| Gemini schema sanitization                  | Strips `$schema`/`$ref`/`additionalProperties` for `gemini-*`/`google-vertex-gemini/*`                                                                                                                                                                                                                                                                                                                      | `auth.loader.fetch` wrap     |
+| Multi-instance                              | Each plugin entry binds to its own `providerId`; closures isolated                                                                                                                                                                                                                                                                                                                                          | factory                      |
+| Config-hook shim                            | OC ≤1.15.5 fallback: writes static catalog into `config.provider[id]` (config hook is the only one that fires in `serve` mode on these versions)                                                                                                                                                                                                                                                            | `config`                     |
+
+## Plugin options
+
+| Option          | Type     | Default                                    | Description                                                |
+| --------------- | -------- | ------------------------------------------ | ---------------------------------------------------------- |
+| `providerId`    | `string` | `"omniroute"`                              | OpenCode provider id; must be unique across plugin entries |
+| `displayName`   | `string` | `"OmniRoute"` or `OmniRoute (<id>)`        | Label in the OC UI                                         |
+| `modelCacheTtl` | `number` | `300000` (5 min)                           | `/v1/models` TTL in ms                                     |
+| `baseURL`       | `string` | resolved from `auth.json` after `/connect` | Override OmniRoute base URL                                |
+| `features`      | `object` | see below                                  | Feature toggles (all opt-in/out, defaults preserve v0.1.0) |
+
+### `features` block
+
+Every field is optional. Defaults mirror v0.1.0 behaviour so existing `opencode.json` files do not need to change.
+
+| Feature               | Type      | Default | What it does                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                 |
+| --------------------- | --------- | ------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| `combos`              | `boolean` | `true`  | Discover `/api/combos` and surface them as pseudo-models with LCD capabilities. Combos are keyed under the `combo/<slug>` namespace and labelled `Combo: <name>` in the model picker so they're distinguishable from raw provider/model pairs.                                                                                                                                                                                                                                                                                                                               |
+| `enrichment`          | `boolean` | `true`  | Pull display names from `/api/pricing/models` AND per-million-token pricing (`input`, `output`, `cached` → `cacheRead`, `cache_creation` → `cacheWrite`) from `/api/pricing`, then overlay both onto the live catalog (so the UI shows `Claude 4.7 Opus` with `cost.input: 5`, `cost.output: 25` instead of raw IDs and zeroed cost).                                                                                                                                                                                                                                        |
+| `compressionMetadata` | `boolean` | `false` | Pull `/api/context/combos` so combo names get tagged with their compression pipeline, e.g. `Combo: claude-primary [rtk🟡 → caveman🟠]`. Intensity tokens render as traffic-light emoji (🟢 lite/minimal · 🟡 standard · 🟠 aggressive/full · 🔴 ultra) so the picker advertises "how compressed" each combo is at a glance.                                                                                                                                                                                                                                                  |
+| `providerTag`         | `boolean` | `true`  | Prepend a short upstream-provider label to the enriched display name with `" - "` separator, so `cc/claude-opus-4-7 → Claude - Claude Opus 4.7` differs visibly from `kr/claude-opus-4-7 → Kiro - Claude Opus 4.7` in the OC TUI model picker. Label resolution: use `/api/pricing/models[<alias>].name` verbatim when ≤8 chars (e.g. `Claude`, `Kiro`, `Codex`, `Qwen`), otherwise fall back to `UPPER(alias)` (e.g. `GitHub Models` → `GHM`, `Gemini-cli` → `GEMINI-CLI`). Idempotent. Combos intentionally skipped (the `Combo: ` prefix already conveys multi-upstream). |
+| `usableOnly`          | `boolean` | `false` | Read `/api/providers` and filter the catalog to providers that have at least one connection with `isActive: true` AND `testStatus: 'active'`. Subtract-filter semantics: providers unknown to BOTH the pricing-models catalog AND the connection table pass through (so synthetic prefixes like `agentrouter/*` survive). On fetch failure the filter is disabled for the refresh — never hides the whole catalog.                                                                                                                                                           |
+| `diskCache`           | `boolean` | `true`  | Persist the last successful `/v1/models` + `/api/combos` + enrichment + connections + compression snapshot to `${OPENCODE_DATA_DIR ?? ~/.local/share/opencode}/plugins/omniroute-<providerId>.json`. On a subsequent cold start where `/v1/models` throws (network down / IP whitelist drop / 5xx) the static block hydrates from the snapshot so OC's model picker survives offline. Soft-fail on read/write — never blocks publishing.                                                                                                                                     |
+| `geminiSanitization`  | `boolean` | `true`  | Strip `$schema`/`$ref`/`additionalProperties` from tool params when the model id matches `gemini`                                                                                                                                                                                                                                                                                                                                                                                                                                                                            |
+| `mcpAutoEmit`         | `boolean` | `false` | Auto-write an `mcp.<providerId>` remote entry into the OC config pointing at `<baseURL>/api/mcp/stream` with the resolved Bearer token                                                                                                                                                                                                                                                                                                                                                                                                                                       |
+| `mcpToken`            | `string`  | _unset_ | Optional separate Bearer for the auto-emitted MCP entry. Falls back to the provider's `apiKey` (from `auth.json`) when unset                                                                                                                                                                                                                                                                                                                                                                                                                                                 |
+| `fetchInterceptor`    | `boolean` | `true`  | Inject `Authorization: Bearer` + default `Content-Type` on every outbound request targeting `baseURL` (suffix-spoof guarded)                                                                                                                                                                                                                                                                                                                                                                                                                                                 |
+
+#### Example — enrichment + compression tags + MCP auto-emit
+
+```jsonc
+{
+  "plugin": [
+    [
+      "@omniroute/opencode-plugin",
+      {
+        "providerId": "omniroute",
+        "baseURL": "https://or.example.com",
+        "features": {
+          "combos": true,
+          "enrichment": true,
+          "compressionMetadata": true,
+          "mcpAutoEmit": true,
+        },
+      },
+    ],
+  ],
+}
+```
+
+With `mcpAutoEmit: true`, the plugin synthesises an `mcp.omniroute` entry equivalent to a manual:
+
+```jsonc
+"mcp": {
+  "omniroute": {
+    "type": "remote",
+    "url": "https://or.example.com/api/mcp/stream",
+    "enabled": true,
+    "headers": { "Authorization": "Bearer <apiKey-from-auth.json>" }
+  }
+}
+```
+
+If you want a narrower-scoped Bearer for MCP (different from the chat/inference key), set `features.mcpToken`. Operator overrides win: if you already set `mcp.omniroute` in `opencode.json`, the plugin will not overwrite it.
+
+#### Example — production-leaning defaults (clean picker, offline resilience)
+
+```jsonc
+{
+  "plugin": [
+    [
+      "@omniroute/opencode-plugin",
+      {
+        "providerId": "omniroute",
+        "baseURL": "https://or.example.com",
+        "features": {
+          "combos": true,
+          "enrichment": true,
+          "compressionMetadata": true,
+          "usableOnly": true,
+          "diskCache": true,
+        },
+      },
+    ],
+  ],
+}
+```
+
+- `usableOnly: true` drops models whose canonical provider has no healthy connection in your OmniRoute instance — your `/models` picker stays focused on what you can actually call.
+- `diskCache: true` (default) writes a snapshot to `${OPENCODE_DATA_DIR}/plugins/omniroute-<providerId>.json` on every healthy refresh. On a cold start where `/v1/models` is unreachable (laptop offline, IP whitelist drop), the snapshot hydrates the static block so OC still shows the catalog instead of a stub.
+- `compressionMetadata: true` annotates combo display names with their pipeline using traffic-light emoji for intensity (e.g. `Combo: claude-primary [rtk🟡 → caveman🟠]`) so the picker advertises which compression each combo applies and how heavy it is at a glance. Palette: 🟢 lite/minimal · 🟡 standard · 🟠 aggressive/full · 🔴 ultra. Unknown intensities fall through to raw text (`[rtk:custom-thing]`) so the plugin never hides a value OmniRoute knows but the plugin doesn't.
+- `providerTag: true` (default) prepends a short upstream-provider label so the picker shows `Claude - Claude Opus 4.7` for `cc/claude-opus-4-7`, `Kiro - Claude Opus 4.7` for `kr/claude-opus-4-7`, and `GHM - GPT 5` for `ghm/gpt-5` (slot.name `GitHub Models` > 8 chars → abbreviated). Critical when the same model id is sold through multiple upstream connections with different cost/auth/rate-limit profiles. Set to `false` to keep the pre-v3.8.3 unsuffixed format.
+
+## Comparison vs `@omniroute/opencode-provider`
+
+[`@omniroute/opencode-provider`](https://github.com/diegosouzapw/OmniRoute/tree/main/%40omniroute/opencode-provider) is the existing config-generator package — it writes a frozen `provider.<id>` block into `opencode.json` at build time. This plugin is the runtime integration.
+
+|                   | `@omniroute/opencode-plugin` (this) | `@omniroute/opencode-provider`    |
+| ----------------- | ----------------------------------- | --------------------------------- |
+| Type              | OC plugin                           | Config generator (CLI/build-time) |
+| Models            | Live from `/v1/models`              | Frozen at scaffold                |
+| Combos            | LCD-aggregated live                 | None                              |
+| Gemini sanitize   | Yes                                 | N/A                               |
+| OC UI integration | `/connect`, `/models`               | None                              |
+| Multi-instance    | Native                              | Manual                            |
+
+Both can coexist; pick the one that fits your environment.
+
+## Requirements
+
+- Node `>=22.22.3` (per `engines.node`); tested on Node 22 and 24.
+- OpenCode: verified end-to-end against `opencode@1.15.5` with `@opencode-ai/plugin@1.15.6`.
+- OC plugin peer (`@opencode-ai/plugin`) `>=1.14.49` for the full feature set (provider hook surfaces models in `/models`). On `<=1.14.48`, the plugin falls back to its `config` hook, writing a static catalog snapshot into `config.provider[id]` so models still appear.
+- The plugin uses the OC v1 plugin shape (`default: { id, server }`) — older OC releases that only walk named exports will reject it. Stay on OC ≥1.15.
+
+## License
+
+MIT. See [LICENSE](./LICENSE).

@omniroute/opencode-plugin/package.json

@@ -0,0 +1,73 @@
+{
+  "name": "@omniroute/opencode-plugin",
+  "version": "0.1.0",
+  "description": "OpenCode plugin for the OmniRoute AI Gateway. Drives dynamic model discovery, /connect auth flow, and multi-instance OmniRoute providers via the official @opencode-ai/plugin contract.",
+  "type": "module",
+  "main": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js"
+    },
+    "./runtime": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js"
+    }
+  },
+  "files": [
+    "dist",
+    "README.md",
+    "LICENSE"
+  ],
+  "scripts": {
+    "build": "tsup",
+    "clean": "rm -rf dist",
+    "test": "node --import tsx/esm --test tests/scaffold.test.ts tests/auth.test.ts tests/options-schema.test.ts tests/multi-instance.test.ts tests/fetch-interceptor.test.ts tests/provider.test.ts tests/gemini-sanitize.test.ts tests/combos.test.ts tests/config-shim.test.ts tests/features.test.ts tests/usable-combo.test.ts tests/disk-snapshot-perms.test.ts tests/fork-features.test.ts tests/auto-combo-context.test.ts",
+    "prepublishOnly": "npm run clean && npm run build && npm test"
+  },
+  "keywords": [
+    "omniroute",
+    "opencode",
+    "opencode-plugin",
+    "ai-sdk",
+    "openai-compatible",
+    "provider",
+    "gemini",
+    "combos",
+    "mcp"
+  ],
+  "author": "OmniRoute contributors",
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/diegosouzapw/OmniRoute.git",
+    "directory": "@omniroute/opencode-plugin"
+  },
+  "bugs": {
+    "url": "https://github.com/diegosouzapw/OmniRoute/issues"
+  },
+  "homepage": "https://github.com/diegosouzapw/OmniRoute/tree/main/%40omniroute/opencode-plugin#readme",
+  "engines": {
+    "node": ">=22.22.3"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "peerDependencies": {
+    "@opencode-ai/plugin": "*"
+  },
+  "dependencies": {
+    "zod": "^4.4.3"
+  },
+  "devDependencies": {
+    "@opencode-ai/plugin": "^1.15.6",
+    "@types/node": "^22.19.19",
+    "tsup": "^8.5.1",
+    "tsx": "^4.22.3",
+    "typescript": "^5.9.3"
+  },
+  "overrides": {
+    "esbuild": "^0.28.1"
+  }
+}

@omniroute/opencode-plugin/src/logger.ts

@@ -0,0 +1,74 @@
+/**
+ * Structured logger for the OmniRoute plugin.
+ *
+ * Levels: error < warn < info < debug
+ * Default: warn (matches current console.warn behavior)
+ * Set via features.logLevel in plugin options.
+ */
+
+export type LogLevel = "error" | "warn" | "info" | "debug";
+
+const LEVEL_ORDER: Record<LogLevel, number> = {
+  error: 0,
+  warn: 1,
+  info: 2,
+  debug: 3,
+};
+
+const TAG = "[omniroute-plugin]";
+
+function shouldLog(current: LogLevel, target: LogLevel): boolean {
+  return LEVEL_ORDER[current] >= LEVEL_ORDER[target];
+}
+
+let _level: LogLevel = "warn";
+
+export function setLogLevel(level: LogLevel): void {
+  _level = level;
+}
+
+export function getLogLevel(): LogLevel {
+  return _level;
+}
+
+function fmt(level: LogLevel, msg: string, tag?: string): string {
+  const prefix = tag ? `${TAG}${tag}` : TAG;
+  return `${prefix} [${level.toUpperCase()}] ${msg}`;
+}
+
+export const logger = {
+  error(msg: string, ...args: unknown[]): void {
+    if (shouldLog(_level, "error")) console.error(fmt("error", msg), ...args);
+  },
+  warn(msg: string, ...args: unknown[]): void {
+    if (shouldLog(_level, "warn")) console.warn(fmt("warn", msg), ...args);
+  },
+  info(msg: string, ...args: unknown[]): void {
+    if (shouldLog(_level, "info")) console.warn(fmt("info", msg), ...args);
+  },
+  debug(msg: string, ...args: unknown[]): void {
+    if (shouldLog(_level, "debug")) console.warn(fmt("debug", msg), ...args);
+  },
+  /** Always emit regardless of level (for critical init breadcrumbs). */
+  always(msg: string, ...args: unknown[]): void {
+    console.warn(TAG, msg, ...args);
+  },
+
+  // ── Tagged child loggers ──────────────────────────────────────────────
+  child(tag: string) {
+    return {
+      error: (msg: string, ...args: unknown[]) =>
+        shouldLog(_level, "error") &&
+        console.error(fmt("error", msg, tag), ...args),
+      warn: (msg: string, ...args: unknown[]) =>
+        shouldLog(_level, "warn") &&
+        console.warn(fmt("warn", msg, tag), ...args),
+      info: (msg: string, ...args: unknown[]) =>
+        shouldLog(_level, "info") &&
+        console.warn(fmt("info", msg, tag), ...args),
+      debug: (msg: string, ...args: unknown[]) =>
+        shouldLog(_level, "debug") &&
+        console.warn(fmt("debug", msg, tag), ...args),
+    };
+  },
+};

@omniroute/opencode-plugin/src/naming.ts

@@ -0,0 +1,301 @@
+/**
+ * Universal model naming template for the OmniRoute plugin.
+ *
+ * Naming pipeline:
+ *   [tag] <provider-label><separator><display-name><suffix>
+ *
+ *   [Free] <provider> - <name> · <budget>     ← free model
+ *   Auto: <variant> (<N>p)                     ← auto combo
+ *   Combo: <name>                              ← DB combo
+ *   <provider> - <name>                        ← regular model
+ */
+
+// ── Constants ────────────────────────────────────────────────────────────
+
+/** Separator between provider label and model display name. */
+export const PROVIDER_TAG_SEPARATOR = " - ";
+
+/** Threshold beyond which providerDisplayName is abbreviated. */
+const PROVIDER_LABEL_MAX_CHARS = 12;
+
+/** Aliases longer than this get title-case instead of UPPER. */
+const ALIAS_UPPER_MAX_CHARS = 5;
+
+// ── Auto Combo Types ─────────────────────────────────────────────────────
+
+export type AutoVariant =
+  | "coding"
+  | "fast"
+  | "cheap"
+  | "offline"
+  | "smart"
+  | "lkgp";
+
+export const AUTO_VARIANTS: AutoVariant[] = [
+  "coding",
+  "fast",
+  "cheap",
+  "offline",
+  "smart",
+  "lkgp",
+];
+
+export const AUTO_VARIANT_DESCRIPTIONS: Record<
+  AutoVariant | "default",
+  string
+> = {
+  default: "Best provider via scoring",
+  coding: "Quality-first for code tasks",
+  fast: "Latency-optimized routing",
+  cheap: "Cost-optimized routing",
+  offline: "Offline-friendly providers",
+  smart: "Quality-first with exploration",
+  lkgp: "Last-Known-Good-Provider routing",
+};
+
+// ── Free Model Types ─────────────────────────────────────────────────────
+
+export type FreeModelFreeType =
+  | "recurring-daily"
+  | "recurring-monthly"
+  | "recurring-credit"
+  | "one-time-initial"
+  | "keyless"
+  | "discontinued";
+
+// ── Provider Label ────────────────────────────────────────────────────────
+
+/**
+ * Title-case a long, lowercase-looking alias.
+ * `antigravity` → `Antigravity`
+ */
+function titleCaseAlias(alias: string): string {
+  if (alias.length === 0) return alias;
+  return alias.charAt(0).toUpperCase() + alias.slice(1).toLowerCase();
+}
+
+/**
+ * Pick the short label for an upstream provider.
+ *
+ * Rules:
+ *   1. Trim `providerDisplayName`. If ≤12 chars → use verbatim.
+ *   2. Alias ≤5 chars → UPPER(alias). Alias >5 → titleCase.
+ *   3. Neither → undefined.
+ */
+export function shortProviderLabel(
+  enrichment:
+    | { providerDisplayName?: string; providerAlias?: string }
+    | undefined,
+): string | undefined {
+  if (!enrichment) return undefined;
+  const raw =
+    typeof enrichment.providerDisplayName === "string"
+      ? enrichment.providerDisplayName.trim()
+      : "";
+  if (raw.length > 0 && raw.length <= PROVIDER_LABEL_MAX_CHARS) return raw;
+  const alias =
+    typeof enrichment.providerAlias === "string"
+      ? enrichment.providerAlias.trim()
+      : "";
+  if (alias.length > 0) {
+    return alias.length <= ALIAS_UPPER_MAX_CHARS
+      ? alias.toUpperCase()
+      : titleCaseAlias(alias);
+  }
+  // Long displayName with no alias to fall back on: keep the long label
+  // rather than dropping the provider prefix entirely.
+  return raw.length > 0 ? raw : undefined;
+}
+
+// ── Free Label ────────────────────────────────────────────────────────────
+
+/**
+ * Normalise display name so free-tier models get a consistent `[Free] ` prefix.
+ *
+ * "GPT-4.1 (Free)"          → "[Free] GPT-4.1"
+ * "DeepSeek V4 Flash Free"  → "[Free] DeepSeek V4 Flash"
+ * "Claude Opus 4.7"         → "Claude Opus 4.7"  (unchanged)
+ */
+export function normaliseFreeLabel(name: string): string {
+  // Bounded whitespace quantifiers ({0,8}/{1,8}) avoid the polynomial-ReDoS
+  // backtracking that unbounded \s* before an anchored \s*$ would allow on
+  // attacker-influenced display names. 8 covers any realistic label spacing.
+  const cleaned = name
+    .replace(/\s{0,8}\(free\)\s{0,8}$/i, "")
+    .replace(/[\s-]{1,8}free\s{0,8}$/i, "")
+    .trim();
+  const wasFree = cleaned.length < name.trim().length;
+  if (!wasFree) return name;
+  return `[Free] ${cleaned}`;
+}
+
+// ── Free Budget Formatting ────────────────────────────────────────────────
+
+function fmtTokens(n: number): string {
+  if (n >= 1e9) return (n / 1e9).toFixed(1).replace(/\.0$/, "") + "B";
+  if (n >= 1e6) return (n / 1e6).toFixed(1).replace(/\.0$/, "") + "M";
+  if (n >= 1e3) return (n / 1e3).toFixed(1).replace(/\.0$/, "") + "K";
+  return String(n);
+}
+
+/**
+ * Format a free model budget into a short human-readable suffix.
+ *
+ * recurring-daily   → "25M tokens/day"
+ * recurring-monthly → "25M tokens/month"
+ * recurring-credit  → "10M credits"
+ * one-time-initial  → "1M credits (one-time)"
+ * keyless           → "(keyless)"
+ * discontinued      → "(discontinued)"
+ */
+export function formatFreeBudget(params: {
+  freeType: FreeModelFreeType;
+  monthlyTokens?: number;
+  creditTokens?: number;
+}): string {
+  const { freeType, monthlyTokens = 0, creditTokens = 0 } = params;
+
+  switch (freeType) {
+    case "recurring-daily":
+      return `${fmtTokens(monthlyTokens)} tokens/day`;
+    case "recurring-monthly":
+      return `${fmtTokens(monthlyTokens)} tokens/month`;
+    case "recurring-credit":
+      return `${fmtTokens(creditTokens)} credits`;
+    case "one-time-initial":
+      return `${fmtTokens(creditTokens)} credits (one-time)`;
+    case "keyless":
+      return "(keyless)";
+    case "discontinued":
+      return "(discontinued)";
+    default:
+      return "";
+  }
+}
+
+// ── Auto Combo Naming ─────────────────────────────────────────────────────
+
+/**
+ * Format auto combo display name.
+ *
+ * "Auto: Coding (4p)"
+ * "Auto: Default (6p)"
+ * "Auto" (no candidate count when unknown)
+ */
+export function formatAutoComboName(
+  variant: AutoVariant | undefined,
+  candidateCount?: number,
+): string {
+  const label = variant
+    ? variant.charAt(0).toUpperCase() + variant.slice(1)
+    : "Default";
+  const count =
+    typeof candidateCount === "number" && candidateCount > 0
+      ? ` (${candidateCount}p)`
+      : "";
+  return `Auto: ${label}${count}`;
+}
+
+/**
+ * Build the model ID for an auto combo entry.
+ * "auto/coding", "auto/fast", "auto" (default).
+ */
+export function autoComboModelId(variant: AutoVariant | undefined): string {
+  return variant ? `auto/${variant}` : "auto";
+}
+
+// ── Universal Display Name Builder ────────────────────────────────────────
+
+export interface ModelDisplayNameParams {
+  /** Raw model ID (e.g. "cc/claude-sonnet-4-6"). */
+  rawId: string;
+  /** Enrichment display name (e.g. "Claude Sonnet 4.6"). */
+  enrichmentName?: string;
+  /** Provider tag enrichment. */
+  providerAlias?: string;
+  /** Human-readable upstream provider label. */
+  providerDisplayName?: string;
+  /** Whether model is free tier. */
+  isFree?: boolean;
+  /** Free model budget info. */
+  freeType?: FreeModelFreeType;
+  /** Monthly token budget (for recurring free models). */
+  monthlyTokens?: number;
+  /** Credit token budget (for credit-based free models). */
+  creditTokens?: number;
+  /** Whether this is a combo entry (skip provider tag). */
+  isCombo?: boolean;
+  /** Whether this is an auto combo entry. */
+  isAutoCombo?: boolean;
+  /** Auto combo variant. */
+  autoVariant?: AutoVariant;
+  /** Auto combo candidate count. */
+  autoCandidateCount?: number;
+}
+
+/**
+ * Build the final display name following the universal template.
+ *
+ * Priority:
+ *   1. Auto combo → "Auto: <variant> (<N>p)"
+ *   2. DB combo → "Combo: <name>"
+ *   3. Free + enrichment + provider tag → "[Free] <label> - <name> · <budget>"
+ *   4. Free + enrichment → "[Free] <name> · <budget>"
+ *   5. Free + raw → "[Free] <rawId> · <budget>"
+ *   6. Enrichment + provider tag → "<label> - <name>"
+ *   7. Enrichment only → "<name>"
+ *   8. Raw fallback → normaliseFreeLabel(rawId)
+ */
+export function buildModelDisplayName(params: ModelDisplayNameParams): string {
+  // Auto combos
+  if (params.isAutoCombo) {
+    return formatAutoComboName(params.autoVariant, params.autoCandidateCount);
+  }
+
+  // Determine base name — strip any existing free suffix first
+  const rawBase =
+    params.enrichmentName && params.enrichmentName.trim().length > 0
+      ? params.enrichmentName
+      : params.rawId;
+  const cleanedBase = rawBase
+    .replace(/\s*\(free\)\s*$/i, "")
+    .replace(/[\s-]+free\s*$/i, "")
+    .trim();
+  const wasFree = cleanedBase.length < rawBase.trim().length;
+  const isFree = !!params.isFree || wasFree;
+
+  let baseName = cleanedBase;
+
+  // Provider tag (skip for combos)
+  if (!params.isCombo) {
+    const label = shortProviderLabel({
+      providerDisplayName: params.providerDisplayName,
+      providerAlias: params.providerAlias,
+    });
+    if (label) {
+      const prefix = `${label}${PROVIDER_TAG_SEPARATOR}`;
+      if (!baseName.startsWith(prefix)) {
+        baseName = `${prefix}${baseName}`;
+      }
+    }
+  }
+
+  // Prepend [Free] if applicable (AFTER provider tag for correct ordering)
+  if (isFree) {
+    baseName = `[Free] ${baseName}`;
+  }
+
+  // Free budget suffix
+  if (isFree && params.freeType) {
+    const budget = formatFreeBudget({
+      freeType: params.freeType,
+      monthlyTokens: params.monthlyTokens,
+      creditTokens: params.creditTokens,
+    });
+    if (budget) {
+      baseName = `${baseName} · ${budget}`;
+    }
+  }
+
+  return baseName;
+}

@omniroute/opencode-plugin/tests/auth.test.ts

@@ -0,0 +1,147 @@
+/**
+ * T-02 auth-hook contract tests.
+ *
+ * Covers the `createOmniRouteAuthHook(opts)` factory and its loader behaviour
+ * against every Auth flavor (`api`, `oauth`, null, empty key). Validates the
+ * multi-instance fix: provider id flows from plugin options, not a module
+ * constant.
+ */
+
+import test from "node:test";
+import assert from "node:assert/strict";
+import { createOmniRouteAuthHook } from "../src/index.js";
+
+test("createOmniRouteAuthHook: default providerId is 'omniroute'", () => {
+  const hook = createOmniRouteAuthHook();
+  assert.equal(hook.provider, "opencode-omniroute");
+});
+
+test("createOmniRouteAuthHook: custom providerId binds to hook.provider (multi-instance)", () => {
+  const hook = createOmniRouteAuthHook({ providerId: "omniroute-preprod" });
+  assert.equal(hook.provider, "opencode-omniroute-preprod");
+});
+
+test("createOmniRouteAuthHook: methods[0] is type 'api' with label including displayName", () => {
+  const hook = createOmniRouteAuthHook();
+  assert.equal(Array.isArray(hook.methods), true);
+  assert.equal(hook.methods.length, 1);
+  const m = hook.methods[0];
+  assert.equal(m.type, "api");
+  assert.equal(m.label, "OmniRoute API Key");
+
+  const custom = createOmniRouteAuthHook({ providerId: "omniroute-preprod" });
+  assert.equal(custom.methods[0].label, "OmniRoute (opencode-omniroute-preprod) API Key");
+});
+
+test("createOmniRouteAuthHook: prompts[0] uses key='apiKey' per @opencode-ai/plugin contract", () => {
+  // NOTE: spec referenced `name: "apiKey"`; the official
+  // @opencode-ai/plugin@1.15.6 prompt shape uses `key` + `message` (no
+  // `name`/`label`/`mask` fields). Asserting against the real type contract.
+  const hook = createOmniRouteAuthHook();
+  const m = hook.methods[0];
+  assert.equal(m.type, "api");
+  // narrow: api method may carry prompts
+  const prompts = "prompts" in m ? m.prompts : undefined;
+  assert.ok(Array.isArray(prompts) && prompts.length === 1, "expected one prompt");
+  const p = prompts![0];
+  assert.equal(p.type, "text");
+  assert.equal((p as { key: string }).key, "apiKey");
+  assert.ok(
+    typeof (p as { message: string }).message === "string" &&
+      (p as { message: string }).message.includes("omniroute"),
+    "prompt message should mention provider id"
+  );
+});
+
+test("loader: valid api auth → {apiKey} when no baseURL option (T-04: fetch omitted)", async () => {
+  // T-04 changed the loader return shape: without a resolvable baseURL the
+  // interceptor cannot gate-keep requests, so the loader falls back to
+  // apiKey-only and the AI-SDK uses its default fetch. See fetch-interceptor
+  // tests for the wired-fetch branches.
+  const hook = createOmniRouteAuthHook();
+  assert.ok(hook.loader, "loader must be defined");
+  const result = await hook.loader!(
+    async () => ({ type: "api", key: "sk-test" }) as never,
+    {} as never
+  );
+  assert.deepEqual(result, { apiKey: "sk-test" });
+});
+
+test("loader: valid api auth → {apiKey, baseURL, fetch} when baseURL option set (T-04)", async () => {
+  const hook = createOmniRouteAuthHook({ baseURL: "https://or.example.com/v1" });
+  const result = await hook.loader!(
+    async () => ({ type: "api", key: "sk-x" }) as never,
+    {} as never
+  );
+  assert.equal((result as { apiKey: string }).apiKey, "sk-x");
+  assert.equal((result as { baseURL: string }).baseURL, "https://or.example.com/v1");
+  assert.equal(
+    typeof (result as { fetch?: unknown }).fetch,
+    "function",
+    "T-04: loader must wire fetch interceptor when baseURL resolves"
+  );
+});
+
+test("loader: features.fetchInterceptor=false AND geminiSanitization=false → no custom fetch (flags honored)", async () => {
+  // Regression: both fetch-layer flags were documented + schema-validated but
+  // silently ignored. Disabling both must fall back to the SDK default fetch.
+  const hook = createOmniRouteAuthHook({
+    baseURL: "https://or.example.com/v1",
+    features: { fetchInterceptor: false, geminiSanitization: false },
+  });
+  const result = await hook.loader!(
+    async () => ({ type: "api", key: "sk-x" }) as never,
+    {} as never
+  );
+  assert.deepEqual(result, { apiKey: "sk-x", baseURL: "https://or.example.com/v1" });
+  assert.equal(
+    (result as { fetch?: unknown }).fetch,
+    undefined,
+    "both flags off must omit the custom fetch"
+  );
+});
+
+test("loader: features.fetchInterceptor=false but geminiSanitization=true → fetch still wired (sanitizer only)", async () => {
+  const hook = createOmniRouteAuthHook({
+    baseURL: "https://or.example.com/v1",
+    features: { fetchInterceptor: false, geminiSanitization: true },
+  });
+  const result = await hook.loader!(
+    async () => ({ type: "api", key: "sk-x" }) as never,
+    {} as never
+  );
+  assert.equal(
+    typeof (result as { fetch?: unknown }).fetch,
+    "function",
+    "geminiSanitization alone must still provide a fetch wrapper"
+  );
+});
+
+test("loader: null/undefined auth → {} (no creds yet, OC surfaces /connect)", async () => {
+  const hook = createOmniRouteAuthHook();
+  const r1 = await hook.loader!(async () => null as never, {} as never);
+  assert.deepEqual(r1, {});
+  const r2 = await hook.loader!(async () => undefined as never, {} as never);
+  assert.deepEqual(r2, {});
+});
+
+test("loader: oauth-flavored auth → {} (wrong method type, ignored)", async () => {
+  const hook = createOmniRouteAuthHook();
+  const result = await hook.loader!(
+    async () =>
+      ({
+        type: "oauth",
+        refresh: "r",
+        access: "a",
+        expires: 0,
+      }) as never,
+    {} as never
+  );
+  assert.deepEqual(result, {});
+});
+
+test("loader: api auth with empty key → {} (empty creds rejected)", async () => {
+  const hook = createOmniRouteAuthHook();
+  const result = await hook.loader!(async () => ({ type: "api", key: "" }) as never, {} as never);
+  assert.deepEqual(result, {});
+});

@omniroute/opencode-plugin/tests/auto-combo-context.test.ts

@@ -0,0 +1,74 @@
+/**
+ * TDD regression — auto combos must never advertise `limit.context: 0`.
+ *
+ * opencode's overflow guard (packages/opencode/src/session/overflow.ts)
+ * short-circuits when `model.limit.context === 0`:
+ *
+ *   if (input.model.limit.context === 0) return false  // never overflow
+ *
+ * so a zero context silently DISABLES opencode's smart auto-compaction for
+ * auto combos. The session then grows unbounded until OmniRoute's
+ * server-side purifyHistory() destructively drops old messages — the
+ * "coding agent keeps forgetting things" bug.
+ *
+ * Fix under test: mapAutoComboToStaticEntry consumes the context_length /
+ * max_output_tokens now served by GET /api/combos/auto, and falls back to a
+ * safe positive default (128000 / 8192) for older servers that do not send
+ * the fields yet.
+ */
+import test from "node:test";
+import assert from "node:assert/strict";
+
+import { mapAutoComboToStaticEntry } from "../src/index.ts";
+import type { OmniRouteRawAutoCombo } from "../src/index.ts";
+
+test("uses server-provided context_length and max_output_tokens", () => {
+  const raw = {
+    id: "auto/coding",
+    name: "Auto Coding",
+    variant: "coding",
+    candidateCount: 5,
+    context_length: 1048576,
+    max_output_tokens: 65536,
+  } as OmniRouteRawAutoCombo;
+
+  const entry = mapAutoComboToStaticEntry(raw);
+  assert.equal(entry.limit?.context, 1048576);
+  assert.equal(entry.limit?.output, 65536);
+});
+
+test("falls back to a safe positive default when the server omits limits (old servers)", () => {
+  const raw = {
+    id: "auto",
+    name: "Auto",
+    candidateCount: 3,
+  } as OmniRouteRawAutoCombo;
+
+  const entry = mapAutoComboToStaticEntry(raw);
+  assert.ok(
+    typeof entry.limit?.context === "number" && entry.limit.context > 0,
+    `context must be a positive number (never 0 — zero disables opencode auto-compaction), got ${entry.limit?.context}`
+  );
+  assert.ok(
+    typeof entry.limit?.output === "number" && entry.limit.output > 0,
+    `output must be a positive number, got ${entry.limit?.output}`
+  );
+});
+
+test("ignores non-positive server values and keeps the safe fallback", () => {
+  const raw = {
+    id: "auto/fast",
+    name: "Auto Fast",
+    variant: "fast",
+    candidateCount: 2,
+    context_length: 0,
+    max_output_tokens: -1,
+  } as OmniRouteRawAutoCombo;
+
+  const entry = mapAutoComboToStaticEntry(raw);
+  assert.ok(
+    typeof entry.limit?.context === "number" && entry.limit.context > 0,
+    "zero/negative server values must not propagate"
+  );
+  assert.ok(typeof entry.limit?.output === "number" && entry.limit.output > 0);
+});

@omniroute/opencode-plugin/tests/combos.test.ts

@@ -0,0 +1,711 @@
+/**
+ * T-05 combo-discovery contract tests.
+ *
+ * Covers:
+ *   - `defaultOmniRouteCombosFetcher(baseURL, apiKey, timeoutMs?)`
+ *     — envelope tolerance (`{combos: [...]}` and bare array), non-2xx errors.
+ *   - `mapComboToModelV2(combo, members, providerId, baseURL)`
+ *     — LCD policy across capabilities, limits, modalities; defensive
+ *       posture on empty members; nice-name preference.
+ *   - `createOmniRouteProviderHook(opts, deps)` extension
+ *     — combos merged into the models map; collision resolution (combo
+ *       wins, warn-once); soft-fail when the combos fetcher throws;
+ *       combos cached + reused under the same TTL key as models.
+ *
+ * Mocking strategy mirrors `provider.test.ts`: both fetchers are
+ * dependency-injected at hook construction, no `fetch` monkey-patch.
+ */
+
+import test from "node:test";
+import assert from "node:assert/strict";
+import {
+  createOmniRouteProviderHook,
+  defaultOmniRouteCombosFetcher,
+  mapComboToModelV2,
+  type OmniRouteCombosFetcher,
+  type OmniRouteModelsFetcher,
+  type OmniRouteRawCombo,
+  type OmniRouteRawModelEntry,
+} from "../src/index.js";
+
+// ────────────────────────────────────────────────────────────────────────────
+// Fixtures
+// ────────────────────────────────────────────────────────────────────────────
+
+const MODEL_PRIMARY: OmniRouteRawModelEntry = {
+  id: "claude-primary",
+  capabilities: {
+    tool_calling: true,
+    reasoning: true,
+    vision: true,
+    thinking: true,
+    temperature: true,
+  },
+  context_length: 200_000,
+  max_output_tokens: 64_000,
+  max_input_tokens: 180_000,
+  input_modalities: ["text", "image"],
+  output_modalities: ["text"],
+};
+
+const MODEL_SECONDARY: OmniRouteRawModelEntry = {
+  id: "claude-secondary",
+  capabilities: {
+    tool_calling: true,
+    reasoning: false,
+    vision: true,
+    thinking: false,
+    temperature: true,
+  },
+  context_length: 100_000,
+  max_output_tokens: 32_000,
+  max_input_tokens: 96_000,
+  input_modalities: ["text", "image"],
+  output_modalities: ["text"],
+};
+
+const MODEL_NO_TOOLS: OmniRouteRawModelEntry = {
+  id: "gemini-3-flash",
+  capabilities: { tool_calling: false, reasoning: false, vision: false, thinking: false },
+  context_length: 1_000_000,
+  max_output_tokens: 8_192,
+  input_modalities: ["text"],
+  output_modalities: ["text"],
+};
+
+const COMBO_CLAUDE_TIER: OmniRouteRawCombo = {
+  id: "combo-claude-tier",
+  name: "Claude Tier",
+  strategy: "priority",
+  models: [
+    { id: "s1", kind: "model", model: "claude-primary", weight: 100 },
+    { id: "s2", kind: "model", model: "claude-secondary", weight: 80 },
+  ],
+};
+
+// ────────────────────────────────────────────────────────────────────────────
+// Helpers
+// ────────────────────────────────────────────────────────────────────────────
+
+function stubModelsFetcher(
+  payload: OmniRouteRawModelEntry[]
+): OmniRouteModelsFetcher & { callCount: () => number } {
+  let n = 0;
+  const f: OmniRouteModelsFetcher = async () => {
+    n++;
+    return payload;
+  };
+  return Object.assign(f, { callCount: () => n });
+}
+
+function stubCombosFetcher(
+  payload: OmniRouteRawCombo[]
+): OmniRouteCombosFetcher & { callCount: () => number; callsBy: () => Array<[string, string]> } {
+  let n = 0;
+  const calls: Array<[string, string]> = [];
+  const f: OmniRouteCombosFetcher = async (baseURL, apiKey) => {
+    n++;
+    calls.push([baseURL, apiKey]);
+    return payload;
+  };
+  return Object.assign(f, {
+    callCount: () => n,
+    callsBy: () => calls,
+  });
+}
+
+function failingCombosFetcher(
+  err = new Error("boom")
+): OmniRouteCombosFetcher & { callCount: () => number } {
+  let n = 0;
+  const f: OmniRouteCombosFetcher = async () => {
+    n++;
+    throw err;
+  };
+  return Object.assign(f, { callCount: () => n });
+}
+
+const apiAuth = (key: string): unknown => ({ type: "api", key });
+
+// Capture console.warn invocations for the duration of a callback, then
+// restore the original. Needed because the collision + soft-fail paths
+// emit warnings we want to assert on.
+async function withWarnCapture<T>(
+  fn: (warnings: Array<{ args: unknown[] }>) => Promise<T>
+): Promise<{ result: T; warnings: Array<{ args: unknown[] }> }> {
+  const original = console.warn;
+  const warnings: Array<{ args: unknown[] }> = [];
+  console.warn = (...args: unknown[]) => {
+    warnings.push({ args });
+  };
+  try {
+    const result = await fn(warnings);
+    return { result, warnings };
+  } finally {
+    console.warn = original;
+  }
+}
+
+// ────────────────────────────────────────────────────────────────────────────
+// defaultOmniRouteCombosFetcher — envelope tolerance + error surfacing
+// ────────────────────────────────────────────────────────────────────────────
+
+test("defaultOmniRouteCombosFetcher: parses {combos:[…]} envelope", async () => {
+  const originalFetch = globalThis.fetch;
+  globalThis.fetch = (async (input: unknown) => {
+    const url = typeof input === "string" ? input : (input as { url: string }).url;
+    assert.equal(url, "https://or.example.com/api/combos");
+    return new Response(
+      JSON.stringify({
+        combos: [
+          { id: "c1", name: "Combo One", strategy: "priority", models: [] },
+          { id: "c2", name: "Combo Two", strategy: "weighted", models: [] },
+        ],
+      }),
+      { status: 200, headers: { "Content-Type": "application/json" } }
+    );
+  }) as typeof fetch;
+  try {
+    const combos = await defaultOmniRouteCombosFetcher("https://or.example.com", "sk-test");
+    assert.equal(combos.length, 2);
+    assert.equal(combos[0].id, "c1");
+    assert.equal(combos[1].id, "c2");
+  } finally {
+    globalThis.fetch = originalFetch;
+  }
+});
+
+test("defaultOmniRouteCombosFetcher: parses bare array envelope", async () => {
+  const originalFetch = globalThis.fetch;
+  globalThis.fetch = (async () => {
+    return new Response(JSON.stringify([{ id: "c1" }, { id: "c2" }, { not_an_id: 42 }]), {
+      status: 200,
+    });
+  }) as typeof fetch;
+  try {
+    const combos = await defaultOmniRouteCombosFetcher("https://or.example.com/v1", "sk-test");
+    // Strip /v1 before /api/combos, AND filter out entries with no string id.
+    assert.equal(combos.length, 2);
+    assert.equal(combos[0].id, "c1");
+    assert.equal(combos[1].id, "c2");
+  } finally {
+    globalThis.fetch = originalFetch;
+  }
+});
+
+test("defaultOmniRouteCombosFetcher: strips trailing /v1 before /api/combos", async () => {
+  const originalFetch = globalThis.fetch;
+  let observedUrl = "";
+  globalThis.fetch = (async (input: unknown) => {
+    observedUrl = typeof input === "string" ? input : (input as { url: string }).url;
+    return new Response(JSON.stringify({ combos: [] }), { status: 200 });
+  }) as typeof fetch;
+  try {
+    await defaultOmniRouteCombosFetcher("https://or.example.com/v1/", "sk-test");
+    assert.equal(observedUrl, "https://or.example.com/api/combos");
+  } finally {
+    globalThis.fetch = originalFetch;
+  }
+});
+
+test("defaultOmniRouteCombosFetcher: throws on non-2xx with status code in message", async () => {
+  const originalFetch = globalThis.fetch;
+  globalThis.fetch = (async () => {
+    return new Response(JSON.stringify({ error: "Invalid token" }), {
+      status: 403,
+      statusText: "Forbidden",
+    });
+  }) as typeof fetch;
+  try {
+    await assert.rejects(
+      async () => {
+        await defaultOmniRouteCombosFetcher("https://or.example.com", "sk-bad");
+      },
+      (err: unknown) => {
+        const msg = err instanceof Error ? err.message : String(err);
+        assert.match(msg, /403/, "status code must appear in message");
+        assert.match(msg, /\/api\/combos/, "url must appear in message");
+        return true;
+      }
+    );
+  } finally {
+    globalThis.fetch = originalFetch;
+  }
+});
+
+test("defaultOmniRouteCombosFetcher: throws when apiKey missing", async () => {
+  await assert.rejects(
+    async () => defaultOmniRouteCombosFetcher("https://or.example.com", ""),
+    /apiKey required/
+  );
+});
+
+test("defaultOmniRouteCombosFetcher: throws when baseURL missing", async () => {
+  await assert.rejects(
+    async () => defaultOmniRouteCombosFetcher("", "sk-test"),
+    /baseURL required/
+  );
+});
+
+// ────────────────────────────────────────────────────────────────────────────
+// mapComboToModelV2 — LCD semantics
+// ────────────────────────────────────────────────────────────────────────────
+
+test("mapComboToModelV2: empty members → capabilities all false (defensive)", () => {
+  const m = mapComboToModelV2(
+    { id: "combo-empty", name: "Empty Combo" },
+    [],
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  assert.equal(m.id, "combo-empty");
+  assert.equal(m.name, "Empty Combo");
+  assert.equal(m.capabilities.temperature, false);
+  assert.equal(m.capabilities.reasoning, false);
+  assert.equal(m.capabilities.attachment, false);
+  assert.equal(m.capabilities.toolcall, false);
+  assert.equal(m.capabilities.input.text, false);
+  assert.equal(m.capabilities.output.text, false);
+  assert.equal(m.limit.context, 0);
+  assert.equal(m.limit.output, 0);
+  assert.equal(m.limit.input, undefined);
+  assert.deepEqual(m.cost, { input: 0, output: 0, cache: { read: 0, write: 0 } });
+});
+
+test("mapComboToModelV2: all members reasoning=true → combo reasoning=true", () => {
+  const m = mapComboToModelV2(
+    { id: "c", models: [] },
+    [
+      MODEL_PRIMARY,
+      {
+        ...MODEL_PRIMARY,
+        id: "p2",
+        capabilities: { ...MODEL_PRIMARY.capabilities, thinking: false, reasoning: true },
+      },
+    ],
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  assert.equal(m.capabilities.reasoning, true);
+});
+
+test("mapComboToModelV2: any member reasoning=false → combo reasoning=false", () => {
+  const m = mapComboToModelV2(
+    { id: "c", models: [] },
+    [MODEL_PRIMARY, MODEL_NO_TOOLS], // gemini-3-flash has reasoning:false, thinking:false
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  assert.equal(m.capabilities.reasoning, false);
+});
+
+test("mapComboToModelV2: limit.context is min of members'", () => {
+  const m = mapComboToModelV2(
+    { id: "c", models: [] },
+    [MODEL_PRIMARY, MODEL_SECONDARY, MODEL_NO_TOOLS],
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  // min(200_000, 100_000, 1_000_000) = 100_000
+  assert.equal(m.limit.context, 100_000);
+  // min(64_000, 32_000, 8_192) = 8_192
+  assert.equal(m.limit.output, 8_192);
+});
+
+test("mapComboToModelV2: limit.input only emitted when EVERY member declares one", () => {
+  const m1 = mapComboToModelV2(
+    { id: "c", models: [] },
+    [MODEL_PRIMARY, MODEL_SECONDARY],
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  // Both declare max_input_tokens → limit.input = min(180000, 96000)
+  assert.equal(m1.limit.input, 96_000);
+
+  const m2 = mapComboToModelV2(
+    { id: "c", models: [] },
+    [MODEL_PRIMARY, MODEL_NO_TOOLS], // gemini-3-flash doesn't declare max_input_tokens
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  assert.equal(m2.limit.input, undefined);
+});
+
+test("mapComboToModelV2: nice name preferred from combo.name", () => {
+  const m1 = mapComboToModelV2(
+    { id: "combo-x", name: "Pretty Name" },
+    [MODEL_PRIMARY],
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  assert.equal(m1.name, "Pretty Name");
+
+  // Falls back to id when name is absent or empty.
+  const m2 = mapComboToModelV2(
+    { id: "combo-y" },
+    [MODEL_PRIMARY],
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  assert.equal(m2.name, "combo-y");
+
+  const m3 = mapComboToModelV2(
+    { id: "combo-z", name: "   " },
+    [MODEL_PRIMARY],
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  assert.equal(m3.name, "combo-z");
+});
+
+test("mapComboToModelV2: attachment AND vision flag both honored across members", () => {
+  // MODEL_PRIMARY: vision=true; MODEL_SECONDARY: vision=true → combo attachment=true
+  const yes = mapComboToModelV2(
+    { id: "c1", models: [] },
+    [MODEL_PRIMARY, MODEL_SECONDARY],
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  assert.equal(yes.capabilities.attachment, true);
+
+  // Add a member with no vision/attachment → AND collapses to false
+  const no = mapComboToModelV2(
+    { id: "c2", models: [] },
+    [MODEL_PRIMARY, MODEL_NO_TOOLS],
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  assert.equal(no.capabilities.attachment, false);
+});
+
+test("mapComboToModelV2: modalities AND'd across members", () => {
+  const m = mapComboToModelV2(
+    { id: "c", models: [] },
+    [MODEL_PRIMARY, MODEL_SECONDARY], // both have text+image
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  assert.equal(m.capabilities.input.text, true);
+  assert.equal(m.capabilities.input.image, true);
+  assert.equal(m.capabilities.input.audio, false);
+
+  // Add a text-only member → image collapses to false.
+  const m2 = mapComboToModelV2(
+    { id: "c", models: [] },
+    [MODEL_PRIMARY, MODEL_NO_TOOLS],
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  assert.equal(m2.capabilities.input.text, true);
+  assert.equal(m2.capabilities.input.image, false);
+});
+
+test("mapComboToModelV2: api block matches providerId + baseURL", () => {
+  const m = mapComboToModelV2(
+    { id: "c" },
+    [MODEL_PRIMARY],
+    "omniroute-preprod",
+    "https://or-preprod.example.com/v1"
+  );
+  assert.equal(m.providerID, "omniroute-preprod");
+  assert.equal(m.api.id, "openai-compatible");
+  assert.equal(m.api.url, "https://or-preprod.example.com/v1");
+  assert.equal(m.api.npm, "@ai-sdk/openai-compatible");
+  assert.equal(m.status, "active");
+});
+
+test("mapComboToModelV2: explicit member temperature=false drops combo temperature=false", () => {
+  const tempFalse: OmniRouteRawModelEntry = {
+    id: "no-temp",
+    capabilities: { tool_calling: true, temperature: false },
+    context_length: 100_000,
+    max_output_tokens: 8_000,
+    input_modalities: ["text"],
+    output_modalities: ["text"],
+  };
+  const m = mapComboToModelV2(
+    { id: "c" },
+    [MODEL_PRIMARY, tempFalse],
+    "omniroute",
+    "https://or.example.com/v1"
+  );
+  assert.equal(m.capabilities.temperature, false);
+});
+
+// ────────────────────────────────────────────────────────────────────────────
+// createOmniRouteProviderHook — combos merge + collision + soft-fail + cache
+// ────────────────────────────────────────────────────────────────────────────
+
+test("models() returns combo entries merged into the map", async () => {
+  const modelsFetcher = stubModelsFetcher([MODEL_PRIMARY, MODEL_SECONDARY, MODEL_NO_TOOLS]);
+  const combosFetcher = stubCombosFetcher([COMBO_CLAUDE_TIER]);
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1" },
+    { fetcher: modelsFetcher, combosFetcher }
+  );
+  const out = await hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+
+  // 3 raw models + 1 combo = 4 entries
+  assert.equal(Object.keys(out).length, 4);
+  assert.ok(out["opencode-omniroute/claude-primary"]);
+  assert.ok(out["opencode-omniroute/claude-secondary"]);
+  assert.ok(out["opencode-omniroute/gemini-3-flash"]);
+  assert.ok(out["opencode-omniroute/claude-tier"]);
+
+  const combo = out["opencode-omniroute/claude-tier"];
+  assert.equal(combo.name, "Claude Tier");
+  assert.equal(combo.providerID, "opencode-omniroute");
+  // LCD over claude-primary (200k, reasoning) + claude-secondary (100k, no reasoning)
+  assert.equal(combo.limit.context, 100_000);
+  assert.equal(combo.capabilities.reasoning, false);
+  assert.equal(combo.capabilities.toolcall, true);
+});
+
+test("models(): combo with unknown member ids degrades to all-false LCD posture", async () => {
+  const modelsFetcher = stubModelsFetcher([MODEL_PRIMARY]); // catalog only has claude-primary
+  const combosFetcher = stubCombosFetcher([
+    {
+      id: "phantom",
+      name: "Phantom Combo",
+      models: [
+        { id: "s1", kind: "model", model: "does-not-exist-1", weight: 50 },
+        { id: "s2", kind: "model", model: "does-not-exist-2", weight: 50 },
+      ],
+    },
+  ]);
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1" },
+    { fetcher: modelsFetcher, combosFetcher }
+  );
+  const out = await hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  assert.ok(out["opencode-omniroute/phantom-combo"]);
+  // With zero resolvable members, LCD = all-false (defensive posture).
+  assert.equal(out["opencode-omniroute/phantom-combo"].capabilities.toolcall, false);
+  assert.equal(out["opencode-omniroute/phantom-combo"].capabilities.reasoning, false);
+  assert.equal(out["opencode-omniroute/phantom-combo"].limit.context, 0);
+});
+
+test("models(): hidden combos are excluded from the map", async () => {
+  const modelsFetcher = stubModelsFetcher([MODEL_PRIMARY]);
+  const combosFetcher = stubCombosFetcher([
+    {
+      id: "visible",
+      name: "Visible",
+      models: [{ id: "s1", kind: "model", model: "claude-primary", weight: 100 }],
+    },
+    {
+      id: "hidden",
+      name: "Hidden",
+      isHidden: true,
+      models: [{ id: "s1", kind: "model", model: "claude-primary", weight: 100 }],
+    },
+  ]);
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1" },
+    { fetcher: modelsFetcher, combosFetcher }
+  );
+  const out = await hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  assert.ok(out["opencode-omniroute/visible"]);
+  assert.ok(!out["opencode-omniroute/hidden"], "hidden combo must be omitted");
+});
+
+test("models(): combo name exactly matches raw model id → raw deleted, raw deleted, no warn", async () => {
+  // Combo.name === raw model id triggers the dedup deletion. This mirrors
+  // the real OmniRoute payload where /v1/models pre-mirrors combos as
+  // no-slash raw entries whose ids match /api/combos friendly names.
+  const colliderCombo: OmniRouteRawCombo = {
+    id: "uuid-collider",
+    name: "claude-primary", // EXACT match to MODEL_PRIMARY.id
+    models: [{ id: "s1", kind: "model", model: "claude-secondary", weight: 100 }],
+  };
+  const modelsFetcher = stubModelsFetcher([MODEL_PRIMARY, MODEL_SECONDARY]);
+  const combosFetcher = stubCombosFetcher([colliderCombo]);
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1" },
+    { fetcher: modelsFetcher, combosFetcher }
+  );
+
+  const { result: out, warnings } = await withWarnCapture(async (_w) => {
+    return hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  });
+
+  // Raw model replaced by combo of the same key; combo now lives at the bare slug.
+  assert.ok(out["opencode-omniroute/claude-primary"], "combo surfaces under prefixed key");
+  assert.equal(out["opencode-omniroute/claude-primary"].name, "claude-primary");
+
+  // No collision warning fires — dedup makes keys disjoint.
+  const collisionWarns = warnings.filter((w) => {
+    const msg = w.args[0];
+    return typeof msg === "string" && msg.includes("collides");
+  });
+  assert.equal(collisionWarns.length, 0, "no collision warn after dedup");
+});
+
+test("models(): two combos with same slug → second gets disambiguator suffix", async () => {
+  // Both combos slug to `claude` — second must get `claude-<id-prefix>`.
+  const combos: OmniRouteRawCombo[] = [
+    {
+      id: "uuid-a",
+      name: "Claude",
+      models: [{ id: "s", kind: "model", model: "claude-primary", weight: 1 }],
+    },
+    {
+      id: "uuid-b",
+      name: "Claude",
+      models: [{ id: "s", kind: "model", model: "claude-secondary", weight: 1 }],
+    },
+  ];
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1" },
+    {
+      fetcher: stubModelsFetcher([MODEL_PRIMARY, MODEL_SECONDARY]),
+      combosFetcher: stubCombosFetcher(combos),
+    }
+  );
+
+  const out = await hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  // First combo gets the bare slug; second gets disambiguated.
+  assert.ok(out["opencode-omniroute/claude"], "first combo at prefixed slug");
+  assert.ok(out["opencode-omniroute/claude-uuid"], "second combo disambiguated by id prefix");
+});
+
+test("models(): combos fetch fails → falls back to models-only, warn emitted, no throw", async () => {
+  const modelsFetcher = stubModelsFetcher([MODEL_PRIMARY, MODEL_SECONDARY]);
+  const combosFetcher = failingCombosFetcher(new Error("ECONNRESET"));
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1" },
+    { fetcher: modelsFetcher, combosFetcher }
+  );
+
+  const { result: out, warnings } = await withWarnCapture(async () => {
+    return hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  });
+
+  // Catalog includes the models but NOT any combo entries.
+  assert.equal(Object.keys(out).length, 2);
+  assert.ok(out["opencode-omniroute/claude-primary"]);
+  assert.ok(out["opencode-omniroute/claude-secondary"]);
+
+  // Soft-fail warning surfaced.
+  const softFail = warnings.find((w) => {
+    const msg = w.args[0];
+    return typeof msg === "string" && msg.includes("combos fetch failed");
+  });
+  assert.ok(softFail, "soft-fail warning must be emitted on combos fetch error");
+  assert.equal(combosFetcher.callCount(), 1);
+});
+
+test("models(): combos cached + reused within TTL (one combo fetch per TTL window)", async () => {
+  const modelsFetcher = stubModelsFetcher([MODEL_PRIMARY, MODEL_SECONDARY]);
+  const combosFetcher = stubCombosFetcher([COMBO_CLAUDE_TIER]);
+  let nowMs = 1_000_000;
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1", modelCacheTtl: 60_000 },
+    { fetcher: modelsFetcher, combosFetcher, now: () => nowMs }
+  );
+
+  await hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  nowMs += 30_000; // half the TTL
+  const second = await hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  assert.equal(combosFetcher.callCount(), 1, "combos fetched only once within TTL");
+  assert.equal(modelsFetcher.callCount(), 1, "models fetched only once within TTL");
+  assert.ok(second["opencode-omniroute/claude-tier"]);
+});
+
+test("models(): combos refetched after TTL expiry (same key as models)", async () => {
+  const modelsFetcher = stubModelsFetcher([MODEL_PRIMARY]);
+  const combosFetcher = stubCombosFetcher([COMBO_CLAUDE_TIER]);
+  let nowMs = 1_000_000;
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1", modelCacheTtl: 60_000 },
+    { fetcher: modelsFetcher, combosFetcher, now: () => nowMs }
+  );
+
+  await hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  nowMs += 60_001;
+  await hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  assert.equal(combosFetcher.callCount(), 2, "combos must refetch past TTL");
+  assert.equal(modelsFetcher.callCount(), 2, "models must refetch past TTL");
+});
+
+test("models(): combos fetcher receives the resolved baseURL + apiKey", async () => {
+  const modelsFetcher = stubModelsFetcher([MODEL_PRIMARY]);
+  const combosFetcher = stubCombosFetcher([COMBO_CLAUDE_TIER]);
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1" },
+    { fetcher: modelsFetcher, combosFetcher }
+  );
+  await hook.models!({} as never, { auth: apiAuth("sk-spy") as never });
+  assert.deepEqual(combosFetcher.callsBy()[0], ["https://or.example.com/v1", "sk-spy"]);
+});
+
+test("models(): nested combo-ref context is the min of nested + raw members", async () => {
+  // Top-level combo MASTER-LIGHT has 1 raw model (claude-primary, 200k)
+  // and 2 combo-refs: OldLLM (8k member) and KIRO (32k member). The OLD
+  // plugin would advertise 200k (only the raw model); the fix should
+  // make it advertise 8k (the bottleneck across the member graph).
+  const modelsFetcher = stubModelsFetcher([
+    MODEL_PRIMARY,
+    {
+      id: "oldllm-member-1",
+      context_length: 8_000,
+      max_output_tokens: 4_000,
+      capabilities: {
+        tool_calling: false,
+        reasoning: false,
+        vision: false,
+        thinking: false,
+        temperature: true,
+      },
+      input_modalities: ["text"],
+      output_modalities: ["text"],
+    },
+    {
+      id: "kiro-member-1",
+      context_length: 32_000,
+      max_output_tokens: 8_000,
+      capabilities: {
+        tool_calling: true,
+        reasoning: false,
+        vision: false,
+        thinking: false,
+        temperature: true,
+      },
+      input_modalities: ["text"],
+      output_modalities: ["text"],
+    },
+  ]);
+  const combosFetcher = stubCombosFetcher([
+    {
+      id: "oldllm",
+      name: "OldLLM",
+      models: [{ id: "s1", kind: "model", model: "oldllm-member-1", weight: 100 }],
+    },
+    {
+      id: "kiro",
+      name: "KIRO",
+      models: [{ id: "s1", kind: "model", model: "kiro-member-1", weight: 100 }],
+    },
+    {
+      id: "master-light",
+      name: "MASTER-LIGHT",
+      models: [
+        { id: "r1", kind: "model", model: "claude-primary", weight: 50 },
+        { id: "r2", kind: "combo-ref", comboName: "OldLLM", weight: 25 },
+        { id: "r3", kind: "combo-ref", comboName: "KIRO", weight: 25 },
+      ],
+    },
+  ]);
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1" },
+    { fetcher: modelsFetcher, combosFetcher }
+  );
+  const out = await hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  const masterLight = out["opencode-omniroute/master-light"];
+  assert.ok(masterLight, "MASTER-LIGHT entry must exist");
+  assert.equal(
+    masterLight.limit.context,
+    8_000,
+    `expected 8_000 (OldLLM bottleneck), got ${masterLight.limit.context}`
+  );
+});

@omniroute/opencode-plugin/tests/disk-snapshot-perms.test.ts

@@ -0,0 +1,66 @@
+/**
+ * Regression test for the disk-snapshot file permissions (release/v3.8.2
+ * review finding C2). The snapshot embeds provider topology + connection
+ * records and lives alongside auth.json (0o600), so it must NOT be readable by
+ * group/other. Before the fix it was written with the default (typically
+ * world-readable 0o644) mode.
+ */
+
+import test from "node:test";
+import assert from "node:assert/strict";
+import fs from "node:fs";
+import os from "node:os";
+import path from "node:path";
+
+import {
+  defaultDiskSnapshotWriter,
+  diskSnapshotPath,
+  type OmniRouteFetchCacheEntry,
+} from "../src/index.js";
+
+function makeEntry(): Omit<OmniRouteFetchCacheEntry, "expiresAt"> {
+  return {
+    rawModels: [],
+    rawCombos: [],
+    rawEnrichment: new Map(),
+    rawCompressionCombos: [],
+    rawConnections: [],
+  };
+}
+
+test("defaultDiskSnapshotWriter writes an owner-only (no group/other) snapshot", async (t) => {
+  // POSIX-only assertion; Windows does not honor numeric file modes.
+  if (process.platform === "win32") {
+    t.skip("file mode semantics are POSIX-only");
+    return;
+  }
+
+  const tmp = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-disk-perms-"));
+  const prevDataDir = process.env.OPENCODE_DATA_DIR;
+  process.env.OPENCODE_DATA_DIR = tmp;
+
+  try {
+    await defaultDiskSnapshotWriter("perm-test", makeEntry());
+
+    const file = diskSnapshotPath("perm-test");
+    assert.ok(fs.existsSync(file), "snapshot file should be written");
+
+    const fileMode = fs.statSync(file).mode & 0o777;
+    assert.equal(
+      fileMode & 0o077,
+      0,
+      `snapshot must not be group/other accessible (got ${fileMode.toString(8)})`
+    );
+
+    const dirMode = fs.statSync(path.dirname(file)).mode & 0o777;
+    assert.equal(
+      dirMode & 0o077,
+      0,
+      `plugins dir must not be group/other accessible (got ${dirMode.toString(8)})`
+    );
+  } finally {
+    if (prevDataDir === undefined) delete process.env.OPENCODE_DATA_DIR;
+    else process.env.OPENCODE_DATA_DIR = prevDataDir;
+    fs.rmSync(tmp, { recursive: true, force: true });
+  }
+});

@omniroute/opencode-plugin/tests/fetch-interceptor.test.ts

@@ -0,0 +1,269 @@
+/**
+ * T-04 fetch-interceptor contract tests.
+ *
+ * Covers `createOmniRouteFetchInterceptor` (URL-prefix gating, header merge,
+ * Content-Type defaulting, input-shape polymorphism) plus the loader
+ * integration that wires it into the AuthHook return shape.
+ *
+ * Strategy: replace `globalThis.fetch` with a closure-based recorder for the
+ * duration of each test (saved-and-restored in try/finally — node:test has
+ * no built-in spy/restore lifecycle). The recorder captures `(input, init)`
+ * as observed by the wrapped global call so we can assert on what was
+ * forwarded after header injection.
+ */
+import test from "node:test";
+import assert from "node:assert/strict";
+import { createOmniRouteAuthHook, createOmniRouteFetchInterceptor } from "../src/index.js";
+
+type FetchCall = { input: Parameters<typeof fetch>[0]; init?: RequestInit };
+
+function installFetchRecorder(response: Response = new Response("ok")) {
+  const calls: FetchCall[] = [];
+  const original = globalThis.fetch;
+  globalThis.fetch = (async (input: any, init?: any) => {
+    calls.push({ input, init });
+    return response;
+  }) as typeof fetch;
+  const restore = () => {
+    globalThis.fetch = original;
+  };
+  return { calls, restore };
+}
+
+const BASE = "https://or.example.com/v1";
+const KEY = "sk-test-fetch";
+
+test("createOmniRouteFetchInterceptor: targets baseURL → Authorization header injected", async () => {
+  const { calls, restore } = installFetchRecorder();
+  try {
+    const f = createOmniRouteFetchInterceptor({ apiKey: KEY, baseURL: BASE });
+    await f(`${BASE}/chat/completions`, {
+      method: "POST",
+      body: JSON.stringify({ x: 1 }),
+    });
+    assert.equal(calls.length, 1);
+    const sent = calls[0]!;
+    const sentHeaders = new Headers((sent.init as RequestInit).headers);
+    assert.equal(sentHeaders.get("Authorization"), `Bearer ${KEY}`);
+  } finally {
+    restore();
+  }
+});
+
+test("createOmniRouteFetchInterceptor: targets baseURL → Authorization OVERRIDES caller-supplied Bearer", async () => {
+  const { calls, restore } = installFetchRecorder();
+  try {
+    const f = createOmniRouteFetchInterceptor({ apiKey: KEY, baseURL: BASE });
+    await f(`${BASE}/chat/completions`, {
+      method: "POST",
+      body: "{}",
+      headers: { Authorization: "Bearer attacker-key" },
+    });
+    const sent = calls[0]!;
+    const sentHeaders = new Headers((sent.init as RequestInit).headers);
+    // We own the apiKey for this provider — caller-supplied Bearer must lose.
+    assert.equal(sentHeaders.get("Authorization"), `Bearer ${KEY}`);
+  } finally {
+    restore();
+  }
+});
+
+test("createOmniRouteFetchInterceptor: targets baseURL + body → Content-Type defaults to application/json", async () => {
+  const { calls, restore } = installFetchRecorder();
+  try {
+    const f = createOmniRouteFetchInterceptor({ apiKey: KEY, baseURL: BASE });
+    await f(`${BASE}/chat/completions`, {
+      method: "POST",
+      body: JSON.stringify({ m: "x" }),
+    });
+    const sent = calls[0]!;
+    const sentHeaders = new Headers((sent.init as RequestInit).headers);
+    assert.equal(sentHeaders.get("Content-Type"), "application/json");
+  } finally {
+    restore();
+  }
+});
+
+test("createOmniRouteFetchInterceptor: caller-set Content-Type is NOT overwritten", async () => {
+  const { calls, restore } = installFetchRecorder();
+  try {
+    const f = createOmniRouteFetchInterceptor({ apiKey: KEY, baseURL: BASE });
+    await f(`${BASE}/v2/whatever`, {
+      method: "POST",
+      body: "raw",
+      headers: { "Content-Type": "text/plain; charset=utf-8" },
+    });
+    const sent = calls[0]!;
+    const sentHeaders = new Headers((sent.init as RequestInit).headers);
+    assert.equal(sentHeaders.get("Content-Type"), "text/plain; charset=utf-8");
+  } finally {
+    restore();
+  }
+});
+
+test("createOmniRouteFetchInterceptor: non-baseURL host → passthrough, no Authorization injected", async () => {
+  const { calls, restore } = installFetchRecorder();
+  try {
+    const f = createOmniRouteFetchInterceptor({ apiKey: KEY, baseURL: BASE });
+    await f("https://third-party.example.org/v1/chat", {
+      method: "POST",
+      body: "{}",
+      headers: { "X-Caller": "yes" },
+    });
+    const sent = calls[0]!;
+    // Init forwarded verbatim — no header injection.
+    const sentHeaders = new Headers((sent.init as RequestInit | undefined)?.headers);
+    assert.equal(sentHeaders.get("Authorization"), null, "MUST NOT leak apiKey");
+    assert.equal(sentHeaders.get("X-Caller"), "yes");
+  } finally {
+    restore();
+  }
+});
+
+test("createOmniRouteFetchInterceptor: refuses suffix-spoof — `${base}-attacker.evil` does NOT match baseURL", async () => {
+  const { calls, restore } = installFetchRecorder();
+  try {
+    const f = createOmniRouteFetchInterceptor({ apiKey: KEY, baseURL: BASE });
+    // baseURL is `https://or.example.com/v1`. A spoofed
+    // `https://or.example.com/v1-attacker.evil/chat` shares the literal prefix
+    // but is NOT under our origin path — must be treated as passthrough.
+    await f("https://or.example.com/v1-attacker.evil/chat", {
+      method: "POST",
+      body: "{}",
+    });
+    const sent = calls[0]!;
+    const sentHeaders = new Headers((sent.init as RequestInit | undefined)?.headers);
+    assert.equal(sentHeaders.get("Authorization"), null);
+  } finally {
+    restore();
+  }
+});
+
+test("createOmniRouteFetchInterceptor: URL object input is handled", async () => {
+  const { calls, restore } = installFetchRecorder();
+  try {
+    const f = createOmniRouteFetchInterceptor({ apiKey: KEY, baseURL: BASE });
+    await f(new URL(`${BASE}/models`), {});
+    const sent = calls[0]!;
+    const sentHeaders = new Headers((sent.init as RequestInit).headers);
+    assert.equal(sentHeaders.get("Authorization"), `Bearer ${KEY}`);
+  } finally {
+    restore();
+  }
+});
+
+test("createOmniRouteFetchInterceptor: Request input is handled (reads .url)", async () => {
+  const { calls, restore } = installFetchRecorder();
+  try {
+    const f = createOmniRouteFetchInterceptor({ apiKey: KEY, baseURL: BASE });
+    const req = new Request(`${BASE}/chat/completions`, {
+      method: "POST",
+      body: JSON.stringify({ a: 1 }),
+      headers: { "X-Caller": "preserved" },
+    });
+    await f(req);
+    const sent = calls[0]!;
+    // The interceptor forwards the original Request as `input` but layers our
+    // headers into the `init`. We assert against the init view since fetch()
+    // resolves headers from init first when both are present.
+    const sentHeaders = new Headers((sent.init as RequestInit).headers);
+    assert.equal(sentHeaders.get("Authorization"), `Bearer ${KEY}`);
+    assert.equal(
+      sentHeaders.get("X-Caller"),
+      "preserved",
+      "Request-attached headers must survive the merge"
+    );
+  } finally {
+    restore();
+  }
+});
+
+test("createOmniRouteFetchInterceptor: trailing slash in baseURL is normalized", async () => {
+  const { calls, restore } = installFetchRecorder();
+  try {
+    const f = createOmniRouteFetchInterceptor({
+      apiKey: KEY,
+      baseURL: `${BASE}////`,
+    });
+    await f(`${BASE}/models`, {});
+    const sent = calls[0]!;
+    const sentHeaders = new Headers((sent.init as RequestInit).headers);
+    assert.equal(sentHeaders.get("Authorization"), `Bearer ${KEY}`);
+  } finally {
+    restore();
+  }
+});
+
+test("createOmniRouteFetchInterceptor: GET without body does NOT set Content-Type", async () => {
+  const { calls, restore } = installFetchRecorder();
+  try {
+    const f = createOmniRouteFetchInterceptor({ apiKey: KEY, baseURL: BASE });
+    await f(`${BASE}/models`); // no init at all
+    const sent = calls[0]!;
+    const sentHeaders = new Headers((sent.init as RequestInit).headers);
+    assert.equal(sentHeaders.get("Authorization"), `Bearer ${KEY}`);
+    assert.equal(
+      sentHeaders.get("Content-Type"),
+      null,
+      "Content-Type should only default when a body exists"
+    );
+  } finally {
+    restore();
+  }
+});
+
+// ----------------------------------------------------------------------------
+// loader integration
+// ----------------------------------------------------------------------------
+
+test("loader: returns fetch fn when apiKey + baseURL both present (via opts)", async () => {
+  const hook = createOmniRouteAuthHook({ baseURL: BASE });
+  const result = await hook.loader!(async () => ({ type: "api", key: KEY }) as never, {} as never);
+  assert.equal((result as { apiKey: string }).apiKey, KEY);
+  assert.equal((result as { baseURL: string }).baseURL, BASE);
+  assert.equal(
+    typeof (result as { fetch?: unknown }).fetch,
+    "function",
+    "loader must wire fetch interceptor when baseURL resolves"
+  );
+});
+
+test("loader: returns fetch fn when baseURL is stashed on the auth credential", async () => {
+  // Some auth backends attach baseURL alongside the key (post-/connect flow).
+  // The loader should pick it up even when plugin opts.baseURL is unset.
+  const hook = createOmniRouteAuthHook();
+  const result = await hook.loader!(
+    async () => ({ type: "api", key: KEY, baseURL: BASE }) as never,
+    {} as never
+  );
+  assert.equal((result as { baseURL?: string }).baseURL, BASE);
+  assert.equal(typeof (result as { fetch?: unknown }).fetch, "function");
+});
+
+test("loader: omits fetch fn when baseURL missing (apiKey-only return)", async () => {
+  const hook = createOmniRouteAuthHook(); // no baseURL opt
+  const result = await hook.loader!(async () => ({ type: "api", key: KEY }) as never, {} as never);
+  // Interceptor needs a baseURL to gate-keep; without one, fall back to
+  // apiKey-only and let the SDK use its default fetch.
+  assert.deepEqual(result, { apiKey: KEY });
+});
+
+test("loader integration: wired interceptor actually injects Bearer when invoked", async () => {
+  // End-to-end: pull the fetch fn out of the loader return and exercise it,
+  // proving the wiring matches the standalone interceptor's contract.
+  const { calls, restore } = installFetchRecorder();
+  try {
+    const hook = createOmniRouteAuthHook({ baseURL: BASE });
+    const result = await hook.loader!(
+      async () => ({ type: "api", key: KEY }) as never,
+      {} as never
+    );
+    const wiredFetch = (result as { fetch: typeof fetch }).fetch;
+    await wiredFetch(`${BASE}/v1/models`, {});
+    assert.equal(calls.length, 1);
+    const sentHeaders = new Headers((calls[0]!.init as RequestInit).headers);
+    assert.equal(sentHeaders.get("Authorization"), `Bearer ${KEY}`);
+  } finally {
+    restore();
+  }
+});

@omniroute/opencode-plugin/tests/fork-features.test.ts

@@ -0,0 +1,291 @@
+/**
+ * Tests for the 3 mrmm-fork features backported to @omniroute/opencode-plugin:
+ *
+ *   1. `normaliseFreeLabel` — free-tier model display names get a consistent
+ *      `[Free] ` prefix instead of trailing "(Free)" or ad-hoc "free" words.
+ *
+ *   2. `resolveApiBlock` — per-provider-prefix API format routing. Anthropic
+ *      prefixes (`cc/`, `claude/`, `anthropic/`, `kiro/`, `kr/`) get the
+ *      Anthropic SDK block; everything else gets OpenAI-compat.
+ *
+ *   3. `debugLog` — JSONL request/response capture, gated by
+ *      `features.debugLog` and togglable at runtime via
+ *      `debugLogEnabled/SetEnabled`.
+ */
+import { test } from "node:test";
+import assert from "node:assert/strict";
+
+import {
+  normaliseFreeLabel,
+  resolveApiBlock,
+  DEFAULT_ANTHROPIC_PREFIXES,
+  ensureV1Suffix,
+  debugLogEnabled,
+  debugLogSetEnabled,
+  debugLogClear,
+  debugLogRead,
+  debugLogAppend,
+  createDebugLoggingFetch,
+  DebugLogEntry,
+} from "../src/index.js";
+
+// ── 1. normaliseFreeLabel ────────────────────────────────────────────────────
+
+test("normaliseFreeLabel: '(Free)' suffix becomes [Free] prefix", () => {
+  assert.equal(normaliseFreeLabel("GPT-4.1 (Free)"), "[Free] GPT-4.1");
+});
+
+test("normaliseFreeLabel: trailing ' Free' word becomes [Free] prefix", () => {
+  assert.equal(
+    normaliseFreeLabel("DeepSeek V4 Flash Free"),
+    "[Free] DeepSeek V4 Flash"
+  );
+});
+
+test("normaliseFreeLabel: trailing '-free' (hyphen) becomes [Free] prefix", () => {
+  assert.equal(normaliseFreeLabel("Llama 4 Scout-free"), "[Free] Llama 4 Scout");
+});
+
+test("normaliseFreeLabel: case-insensitive (FREE, Free, free all match)", () => {
+  assert.equal(normaliseFreeLabel("Model A FREE"), "[Free] Model A");
+  assert.equal(normaliseFreeLabel("Model A free"), "[Free] Model A");
+  assert.equal(normaliseFreeLabel("Model A Free"), "[Free] Model A");
+});
+
+test("normaliseFreeLabel: names without 'free' pass through unchanged", () => {
+  assert.equal(normaliseFreeLabel("Claude 4.7 Opus"), "Claude 4.7 Opus");
+  assert.equal(normaliseFreeLabel("GPT-5"), "GPT-5");
+});
+
+test("normaliseFreeLabel: 'free' in the middle of a name is NOT rewritten", () => {
+  // Only trailing/standalone "free" markers count; embedded "freedom" stays
+  assert.equal(
+    normaliseFreeLabel("Freedom Model"),
+    "Freedom Model"
+  );
+});
+
+test("normaliseFreeLabel: empty / whitespace-only inputs are handled", () => {
+  // Empty input returns empty; pure whitespace input passes through (no Free marker)
+  assert.equal(normaliseFreeLabel(""), "");
+  assert.equal(normaliseFreeLabel("   "), "   ");
+});
+
+// ── 2. resolveApiBlock ───────────────────────────────────────────────────────
+
+test("resolveApiBlock: cc/* models get the Anthropic SDK block (no /v1)", () => {
+  const block = resolveApiBlock("cc/claude-opus-4-7", "https://api.example.com");
+  assert.equal(block.id, "anthropic");
+  assert.equal(block.npm, "@ai-sdk/anthropic");
+  assert.equal(block.url, "https://api.example.com"); // NO /v1 suffix
+});
+
+test("resolveApiBlock: claude/*, anthropic/*, kiro/*, kr/* all route to Anthropic", () => {
+  for (const id of [
+    "claude/claude-opus-4-7",
+    "anthropic/claude-sonnet-4",
+    "kiro/claude-sonnet-4-5",
+    "kr/claude-opus-4-6",
+  ]) {
+    const block = resolveApiBlock(id, "https://api.example.com");
+    assert.equal(block.id, "anthropic", `${id} should route to Anthropic`);
+    assert.equal(block.npm, "@ai-sdk/anthropic");
+  }
+});
+
+test("resolveApiBlock: non-Anthropic models get OpenAI-compat with /v1", () => {
+  const block = resolveApiBlock("gpt-4o", "https://api.example.com");
+  assert.equal(block.id, "openai-compatible");
+  assert.equal(block.npm, "@ai-sdk/openai-compatible");
+  assert.equal(block.url, "https://api.example.com/v1");
+});
+
+test("resolveApiBlock: user can override anthropicPrefixes to add custom prefixes", () => {
+  const block = resolveApiBlock("myproxy/claude-opus", "https://api.example.com", {
+    anthropicPrefixes: ["myproxy"],
+  });
+  assert.equal(block.id, "anthropic");
+  assert.equal(block.npm, "@ai-sdk/anthropic");
+});
+
+test("resolveApiBlock: empty anthropicPrefixes forces OpenAI-compat for everything", () => {
+  const block = resolveApiBlock("cc/claude-opus", "https://api.example.com", {
+    anthropicPrefixes: [],
+  });
+  assert.equal(block.id, "openai-compatible");
+});
+
+test("resolveApiBlock: baseURL that already ends in /v1 is not double-suffixed (OpenAI path)", () => {
+  const block = resolveApiBlock("gpt-4o", "https://api.example.com/v1");
+  assert.equal(block.url, "https://api.example.com/v1"); // idempotent
+});
+
+test("resolveApiBlock: model id without '/' uses the id as prefix", () => {
+  const block = resolveApiBlock("claude-opus-4-7", "https://api.example.com");
+  // The whole id is the prefix, which doesn't match "cc"/"claude" etc.
+  // So it falls through to OpenAI-compat.
+  assert.equal(block.id, "openai-compatible");
+});
+
+test("DEFAULT_ANTHROPIC_PREFIXES: contains the canonical Anthropic aliases", () => {
+  assert.deepEqual(DEFAULT_ANTHROPIC_PREFIXES, [
+    "cc",
+    "claude",
+    "anthropic",
+    "kiro",
+    "kr",
+  ]);
+});
+
+test("ensureV1Suffix: idempotent for URLs that already end in /v1", () => {
+  assert.equal(ensureV1Suffix("https://api.example.com/v1"), "https://api.example.com/v1");
+  assert.equal(
+    ensureV1Suffix("https://api.example.com/v1/"),
+    "https://api.example.com/v1" // trailing slash is stripped
+  );
+});
+
+test("ensureV1Suffix: appends /v1 when missing", () => {
+  assert.equal(ensureV1Suffix("https://api.example.com"), "https://api.example.com/v1");
+  assert.equal(ensureV1Suffix("https://api.example.com/"), "https://api.example.com/v1");
+});
+
+// ── 3. debugLog ──────────────────────────────────────────────────────────────
+
+test("debugLog: default state is disabled", () => {
+  debugLogClear("test-provider-disabled-default");
+  assert.equal(debugLogEnabled("test-provider-disabled-default"), false);
+});
+
+test("debugLogSetEnabled + debugLogEnabled: roundtrip", () => {
+  debugLogSetEnabled("test-provider-toggle", true);
+  assert.equal(debugLogEnabled("test-provider-toggle"), true);
+  debugLogSetEnabled("test-provider-toggle", false);
+  assert.equal(debugLogEnabled("test-provider-toggle"), false);
+});
+
+test("debugLogAppend + debugLogRead: roundtrip preserves entry shape", () => {
+  const providerId = "test-provider-readroundtrip";
+  debugLogClear(providerId);
+  const entry: DebugLogEntry = {
+    reqId: "req-1",
+    providerId,
+    ts: 1700000000000,
+    url: "https://api.example.com/v1/chat",
+    method: "POST",
+    reqHeaders: { "content-type": "application/json" },
+    reqBody: { model: "gpt-4o", messages: [] },
+    resStatus: 200,
+    resHeaders: { "content-type": "application/json" },
+    resBody: { choices: [] },
+    durationMs: 42,
+  };
+  debugLogAppend(entry);
+  const read = debugLogRead(providerId, 10);
+  assert.equal(read.length, 1);
+  assert.deepEqual(read[0], entry);
+});
+
+test("createDebugLoggingFetch: passes through when disabled", async () => {
+  const providerId = "test-provider-passthrough";
+  debugLogClear(providerId);
+  debugLogSetEnabled(providerId, false);
+  const calls: unknown[] = [];
+  const inner: typeof fetch = async (input) => {
+    calls.push(input);
+    return new Response("ok", { status: 200 });
+  };
+  const wrapped = createDebugLoggingFetch(inner, providerId, false);
+  const res = await wrapped("https://api.example.com/v1/chat");
+  assert.equal(res.status, 200);
+  assert.equal(calls.length, 1);
+  // No log entry should be written when disabled
+  assert.equal(debugLogRead(providerId).length, 0);
+});
+
+test("createDebugLoggingFetch: captures request/response when enabled", async () => {
+  const providerId = "test-provider-captures";
+  debugLogClear(providerId);
+  const inner: typeof fetch = async () =>
+    new Response(JSON.stringify({ ok: true }), {
+      status: 200,
+      headers: { "content-type": "application/json" },
+    });
+  const wrapped = createDebugLoggingFetch(inner, providerId, true);
+  const res = await wrapped("https://api.example.com/v1/chat", {
+    method: "POST",
+    headers: { "content-type": "application/json" },
+    body: JSON.stringify({ model: "gpt-4o" }),
+  });
+  assert.equal(res.status, 200);
+  const entries = debugLogRead(providerId);
+  assert.equal(entries.length, 1);
+  assert.equal(entries[0].method, "POST");
+  assert.equal(entries[0].resStatus, 200);
+  assert.equal(entries[0].url, "https://api.example.com/v1/chat");
+  assert.deepEqual(entries[0].reqBody, { model: "gpt-4o" });
+});
+
+test("createDebugLoggingFetch: records error without crashing the wrapped fetch", async () => {
+  const providerId = "test-provider-error";
+  debugLogClear(providerId);
+  const inner: typeof fetch = async () => {
+    throw new Error("network down");
+  };
+  const wrapped = createDebugLoggingFetch(inner, providerId, true);
+  await assert.rejects(wrapped("https://api.example.com/v1/chat"), /network down/);
+  const entries = debugLogRead(providerId);
+  assert.equal(entries.length, 1);
+  assert.equal(entries[0].resStatus, null);
+  assert.equal(entries[0].error, "network down");
+});
+
+// ── Regression tests for the 3 HIGH-priority bot review fixes ───────────────
+
+test("createDebugLoggingFetch: URL instance input is captured (not 'undefined')", async () => {
+  const providerId = "test-provider-url-input";
+  debugLogClear(providerId);
+  const inner: typeof fetch = async () =>
+    new Response("ok", { status: 200 });
+  const wrapped = createDebugLoggingFetch(inner, providerId, true);
+  await wrapped(new URL("https://api.example.com/v1/chat"));
+  const entries = debugLogRead(providerId);
+  assert.equal(entries.length, 1);
+  assert.equal(entries[0].url, "https://api.example.com/v1/chat");
+  assert.notEqual(entries[0].url, undefined);
+});
+
+test("createDebugLoggingFetch: Request object input captures URL and headers", async () => {
+  const providerId = "test-provider-request-input";
+  debugLogClear(providerId);
+  const inner: typeof fetch = async () =>
+    new Response("ok", { status: 200 });
+  const wrapped = createDebugLoggingFetch(inner, providerId, true);
+  const req = new Request("https://api.example.com/v1/chat", {
+    method: "POST",
+    headers: { "x-test": "yes" },
+  });
+  await wrapped(req);
+  const entries = debugLogRead(providerId);
+  assert.equal(entries.length, 1);
+  assert.equal(entries[0].url, "https://api.example.com/v1/chat");
+  assert.equal(entries[0].reqHeaders["x-test"], "yes");
+});
+
+test("createDebugLoggingFetch: SSE response is NOT buffered (resBody is the stream marker)", async () => {
+  const providerId = "test-provider-sse";
+  debugLogClear(providerId);
+  const inner: typeof fetch = async () =>
+    new Response("data: hello\n\n", {
+      status: 200,
+      headers: { "content-type": "text/event-stream" },
+    });
+  const wrapped = createDebugLoggingFetch(inner, providerId, true);
+  const res = await wrapped("https://api.example.com/v1/stream");
+  // The response body must remain readable downstream
+  const txt = await res.text();
+  assert.equal(txt, "data: hello\n\n");
+  const entries = debugLogRead(providerId);
+  assert.equal(entries.length, 1);
+  assert.equal(entries[0].resBody, "[stream]", "SSE responses must not be buffered");
+});

@omniroute/opencode-plugin/tests/gemini-sanitize.test.ts

@@ -0,0 +1,410 @@
+/**
+ * T-06 Gemini tool-schema sanitisation contract tests.
+ *
+ * Three layers under test:
+ *   1. `sanitizeGeminiToolSchemas` — pure function; key stripping + clone
+ *      semantics on chat-completion + Responses-API shapes.
+ *   2. `shouldSanitizeForGemini` — model-string detection (liberal).
+ *   3. `createGeminiSanitizingFetch` — wrapper composition; URL gating,
+ *      body-shape polymorphism, streaming-body bypass, fail-open behaviour,
+ *      composition with the T-04 Bearer interceptor.
+ *
+ * Strategy: same posture as fetch-interceptor.test.ts — install a
+ * closure-based fetch recorder; assert on the `(input, init)` observed by
+ * the inner fetch after the sanitising wrapper has had its say.
+ */
+import test from "node:test";
+import assert from "node:assert/strict";
+import {
+  __resetGeminiStreamingWarning,
+  createGeminiSanitizingFetch,
+  createOmniRouteFetchInterceptor,
+  sanitizeGeminiToolSchemas,
+  shouldSanitizeForGemini,
+} from "../src/index.js";
+
+// ────────────────────────────────────────────────────────────────────────────
+// Helpers
+// ────────────────────────────────────────────────────────────────────────────
+
+type FetchCall = { input: Parameters<typeof fetch>[0]; init?: RequestInit };
+
+function recorder(response: Response = new Response("ok")): {
+  fn: typeof fetch;
+  calls: FetchCall[];
+} {
+  const calls: FetchCall[] = [];
+  const fn = (async (input: any, init?: any) => {
+    calls.push({ input, init });
+    return response;
+  }) as typeof fetch;
+  return { fn, calls };
+}
+
+function bodyAsRecord(init: RequestInit | undefined): Record<string, unknown> {
+  const b = init?.body;
+  if (typeof b !== "string") {
+    throw new Error(`expected string body, got ${typeof b}`);
+  }
+  return JSON.parse(b) as Record<string, unknown>;
+}
+
+// Sample tool payloads — small enough to inline, big enough to cover
+// chat-completion + Responses-API + nested properties.
+
+function chatCompletionsWithDollarSchema(): Record<string, unknown> {
+  return {
+    model: "gemini-2.5-pro",
+    tools: [
+      {
+        type: "function",
+        function: {
+          name: "search",
+          parameters: {
+            $schema: "http://json-schema.org/draft-07/schema#",
+            type: "object",
+            additionalProperties: false,
+            properties: {
+              q: { type: "string" },
+            },
+            required: ["q"],
+          },
+        },
+      },
+    ],
+  };
+}
+
+function responsesApiWithRef(): Record<string, unknown> {
+  return {
+    model: "gemini-2.5-flash",
+    tools: [
+      {
+        type: "function",
+        name: "lookup",
+        input_schema: {
+          type: "object",
+          $ref: "#/definitions/Lookup",
+          properties: {
+            id: { type: "string", ref: "Id" },
+          },
+        },
+      },
+    ],
+  };
+}
+
+function nestedPropertiesPayload(): Record<string, unknown> {
+  return {
+    model: "gemini-pro",
+    tools: [
+      {
+        type: "function",
+        function: {
+          name: "deep",
+          parameters: {
+            type: "object",
+            properties: {
+              outer: {
+                type: "object",
+                $schema: "http://json-schema.org/draft-07/schema#",
+                properties: {
+                  inner: {
+                    type: "object",
+                    additionalProperties: true,
+                    $ref: "#/inner",
+                    properties: {
+                      leaf: { type: "string" },
+                    },
+                  },
+                },
+              },
+            },
+          },
+        },
+      },
+    ],
+  };
+}
+
+// ────────────────────────────────────────────────────────────────────────────
+// sanitizeGeminiToolSchemas — pure function
+// ────────────────────────────────────────────────────────────────────────────
+
+test("sanitizeGeminiToolSchemas: strips $schema from top-level", () => {
+  const input = {
+    model: "gemini-2.5-pro",
+    $schema: "http://json-schema.org/draft-07/schema#",
+    tools: [],
+  };
+  const out = sanitizeGeminiToolSchemas(input) as Record<string, unknown>;
+  assert.equal(out.$schema, undefined);
+  assert.equal(out.model, "gemini-2.5-pro");
+});
+
+test("sanitizeGeminiToolSchemas: strips $ref + additionalProperties from tools[].function.parameters", () => {
+  const input = chatCompletionsWithDollarSchema();
+  const out = sanitizeGeminiToolSchemas(input) as Record<string, unknown>;
+  const params = (out.tools as Array<{ function: { parameters: Record<string, unknown> } }>)[0]!
+    .function.parameters;
+  assert.equal(params.$schema, undefined);
+  assert.equal(params.additionalProperties, undefined);
+  // Untouched keys survive.
+  assert.equal(params.type, "object");
+  assert.deepEqual(params.required, ["q"]);
+});
+
+test("sanitizeGeminiToolSchemas: strips nested $schema from properties.x.properties.y", () => {
+  const input = nestedPropertiesPayload();
+  const out = sanitizeGeminiToolSchemas(input) as Record<string, unknown>;
+  const params = (out.tools as Array<{ function: { parameters: Record<string, unknown> } }>)[0]!
+    .function.parameters;
+  const outer = (params.properties as Record<string, Record<string, unknown>>).outer!;
+  const inner = (outer.properties as Record<string, Record<string, unknown>>).inner!;
+  assert.equal(outer.$schema, undefined);
+  assert.equal(inner.$ref, undefined);
+  assert.equal(inner.additionalProperties, undefined);
+  // Leaf still intact.
+  assert.deepEqual(inner.properties, { leaf: { type: "string" } });
+});
+
+test("sanitizeGeminiToolSchemas: handles Responses-API tools[].input_schema shape", () => {
+  const input = responsesApiWithRef();
+  const out = sanitizeGeminiToolSchemas(input) as Record<string, unknown>;
+  const inputSchema = (out.tools as Array<{ input_schema: Record<string, unknown> }>)[0]!
+    .input_schema;
+  assert.equal(inputSchema.$ref, undefined);
+  // Nested `ref` (lowercase) also stripped.
+  const props = inputSchema.properties as Record<string, Record<string, unknown>>;
+  assert.equal(props.id!.ref, undefined);
+  assert.equal(props.id!.type, "string");
+});
+
+test("sanitizeGeminiToolSchemas: leaves payload without tools untouched", () => {
+  const input = { model: "gemini-2.5-pro", messages: [{ role: "user", content: "hi" }] };
+  const out = sanitizeGeminiToolSchemas(input) as Record<string, unknown>;
+  assert.deepEqual(out, input);
+});
+
+test("sanitizeGeminiToolSchemas: does not mutate input (returned object is distinct)", () => {
+  const input = chatCompletionsWithDollarSchema();
+  const beforeJson = JSON.stringify(input);
+  const out = sanitizeGeminiToolSchemas(input);
+  // Input bit-identical to its pre-sanitise serialisation.
+  assert.equal(JSON.stringify(input), beforeJson);
+  // Output is a different reference.
+  assert.notEqual(out, input);
+});
+
+// ────────────────────────────────────────────────────────────────────────────
+// shouldSanitizeForGemini — detection
+// ────────────────────────────────────────────────────────────────────────────
+
+test("shouldSanitizeForGemini: gemini-2.5-pro → true", () => {
+  assert.equal(shouldSanitizeForGemini({ model: "gemini-2.5-pro" }), true);
+});
+
+test("shouldSanitizeForGemini: models/gemini-pro → true", () => {
+  assert.equal(shouldSanitizeForGemini({ model: "models/gemini-pro" }), true);
+});
+
+test("shouldSanitizeForGemini: google-vertex/gemini-1.5-flash → true", () => {
+  assert.equal(shouldSanitizeForGemini({ model: "google-vertex/gemini-1.5-flash" }), true);
+});
+
+test("shouldSanitizeForGemini: gemini-cli/gemini-2.5-pro → true (real OmniRoute alias)", () => {
+  assert.equal(shouldSanitizeForGemini({ model: "gemini-cli/gemini-2.5-pro" }), true);
+});
+
+test("shouldSanitizeForGemini: claude-sonnet-4 → false", () => {
+  assert.equal(shouldSanitizeForGemini({ model: "claude-sonnet-4" }), false);
+});
+
+test("shouldSanitizeForGemini: payload.model missing → false", () => {
+  assert.equal(shouldSanitizeForGemini({ messages: [] }), false);
+});
+
+test("shouldSanitizeForGemini: payload is null → false", () => {
+  assert.equal(shouldSanitizeForGemini(null), false);
+});
+
+test("shouldSanitizeForGemini: payload.model is non-string → false", () => {
+  assert.equal(shouldSanitizeForGemini({ model: 42 }), false);
+});
+
+// ────────────────────────────────────────────────────────────────────────────
+// createGeminiSanitizingFetch — wrapper
+// ────────────────────────────────────────────────────────────────────────────
+
+const URL_CHAT = "https://or.example.com/v1/chat/completions";
+const URL_RESPONSES = "https://or.example.com/v1/responses";
+const URL_MODELS = "https://or.example.com/v1/models";
+
+test("createGeminiSanitizingFetch: gemini model + chat/completions → tool schemas stripped before forward", async () => {
+  const rec = recorder();
+  const wrapped = createGeminiSanitizingFetch(rec.fn);
+  await wrapped(URL_CHAT, {
+    method: "POST",
+    body: JSON.stringify(chatCompletionsWithDollarSchema()),
+  });
+  assert.equal(rec.calls.length, 1);
+  const forwarded = bodyAsRecord(rec.calls[0]!.init);
+  const params = (
+    forwarded.tools as Array<{ function: { parameters: Record<string, unknown> } }>
+  )[0]!.function.parameters;
+  assert.equal(params.$schema, undefined);
+  assert.equal(params.additionalProperties, undefined);
+});
+
+test("createGeminiSanitizingFetch: non-gemini model + chat/completions → body passed through unchanged", async () => {
+  const rec = recorder();
+  const wrapped = createGeminiSanitizingFetch(rec.fn);
+  const originalBody = JSON.stringify({
+    model: "claude-sonnet-4",
+    tools: [
+      {
+        type: "function",
+        function: {
+          name: "x",
+          parameters: { $schema: "keep-me", type: "object" },
+        },
+      },
+    ],
+  });
+  await wrapped(URL_CHAT, { method: "POST", body: originalBody });
+  // Identity check on body — wrapper must NOT mutate non-Gemini payloads.
+  assert.equal(rec.calls[0]!.init!.body, originalBody);
+});
+
+test("createGeminiSanitizingFetch: gemini model + /v1/models (non-completion endpoint) → body passed through unchanged", async () => {
+  const rec = recorder();
+  const wrapped = createGeminiSanitizingFetch(rec.fn);
+  // GET /v1/models has no body in production; assert that even if a caller
+  // attached a Gemini-shaped body to a non-completion URL, the wrapper
+  // doesn't touch it.
+  const body = JSON.stringify(chatCompletionsWithDollarSchema());
+  await wrapped(URL_MODELS, { method: "POST", body });
+  assert.equal(rec.calls[0]!.init!.body, body);
+});
+
+test("createGeminiSanitizingFetch: gemini model + /responses endpoint → input_schema stripped", async () => {
+  const rec = recorder();
+  const wrapped = createGeminiSanitizingFetch(rec.fn);
+  await wrapped(URL_RESPONSES, {
+    method: "POST",
+    body: JSON.stringify(responsesApiWithRef()),
+  });
+  const forwarded = bodyAsRecord(rec.calls[0]!.init);
+  const schema = (forwarded.tools as Array<{ input_schema: Record<string, unknown> }>)[0]!
+    .input_schema;
+  assert.equal(schema.$ref, undefined);
+});
+
+test("createGeminiSanitizingFetch: gemini model + Request input with body → tool schemas stripped", async () => {
+  const rec = recorder();
+  const wrapped = createGeminiSanitizingFetch(rec.fn);
+  const req = new Request(URL_CHAT, {
+    method: "POST",
+    body: JSON.stringify(chatCompletionsWithDollarSchema()),
+    headers: { "Content-Type": "application/json" },
+  });
+  await wrapped(req);
+  const forwarded = bodyAsRecord(rec.calls[0]!.init);
+  const params = (
+    forwarded.tools as Array<{ function: { parameters: Record<string, unknown> } }>
+  )[0]!.function.parameters;
+  assert.equal(params.$schema, undefined);
+});
+
+test("createGeminiSanitizingFetch: gemini model + ReadableStream body → skipped + warn emitted once", async () => {
+  __resetGeminiStreamingWarning();
+  const rec = recorder();
+  const wrapped = createGeminiSanitizingFetch(rec.fn);
+
+  // Capture console.warn for the duration of this test.
+  const warnings: string[] = [];
+  const originalWarn = console.warn;
+  console.warn = (...args: unknown[]) => {
+    warnings.push(args.map(String).join(" "));
+  };
+
+  try {
+    const stream1 = new ReadableStream({
+      start(controller) {
+        controller.enqueue(new TextEncoder().encode("{}"));
+        controller.close();
+      },
+    });
+    const stream2 = new ReadableStream({
+      start(controller) {
+        controller.enqueue(new TextEncoder().encode("{}"));
+        controller.close();
+      },
+    });
+    // Two streaming calls — only one warn expected.
+    await wrapped(URL_CHAT, { method: "POST", body: stream1 });
+    await wrapped(URL_CHAT, { method: "POST", body: stream2 });
+  } finally {
+    console.warn = originalWarn;
+  }
+
+  // Both calls forwarded to inner fetch with their streams intact.
+  assert.equal(rec.calls.length, 2);
+  // ONE warning total — one-shot latch held.
+  assert.equal(warnings.length, 1);
+  assert.match(warnings[0]!, /streaming Request body, skipping schema strip/);
+});
+
+test("createGeminiSanitizingFetch: invalid JSON body → pass through, no throw", async () => {
+  const rec = recorder();
+  const wrapped = createGeminiSanitizingFetch(rec.fn);
+  // Garbage body must not crash the wrapper.
+  await wrapped(URL_CHAT, { method: "POST", body: "this is not json{{" });
+  assert.equal(rec.calls.length, 1);
+  assert.equal(rec.calls[0]!.init!.body, "this is not json{{");
+});
+
+test("createGeminiSanitizingFetch: empty body → pass through unchanged", async () => {
+  const rec = recorder();
+  const wrapped = createGeminiSanitizingFetch(rec.fn);
+  await wrapped(URL_CHAT, { method: "POST" });
+  assert.equal(rec.calls.length, 1);
+});
+
+test("createGeminiSanitizingFetch: composes correctly with createOmniRouteFetchInterceptor (Bearer + sanitization)", async () => {
+  // Save and replace globalThis.fetch — the Bearer interceptor calls global
+  // fetch when the URL targets its baseURL.
+  const originalFetch = globalThis.fetch;
+  const observed: FetchCall[] = [];
+  globalThis.fetch = (async (input: any, init?: any) => {
+    observed.push({ input, init });
+    return new Response("ok");
+  }) as typeof fetch;
+
+  try {
+    const composed = createGeminiSanitizingFetch(
+      createOmniRouteFetchInterceptor({
+        apiKey: "sk-test",
+        baseURL: "https://or.example.com/v1",
+      })
+    );
+    await composed(URL_CHAT, {
+      method: "POST",
+      body: JSON.stringify(chatCompletionsWithDollarSchema()),
+    });
+
+    assert.equal(observed.length, 1);
+    // Bearer injected (header concern).
+    const sentHeaders = new Headers((observed[0]!.init as RequestInit).headers);
+    assert.equal(sentHeaders.get("Authorization"), "Bearer sk-test");
+    // Schema sanitised (body concern).
+    const forwarded = bodyAsRecord(observed[0]!.init);
+    const params = (
+      forwarded.tools as Array<{ function: { parameters: Record<string, unknown> } }>
+    )[0]!.function.parameters;
+    assert.equal(params.$schema, undefined);
+    assert.equal(params.additionalProperties, undefined);
+  } finally {
+    globalThis.fetch = originalFetch;
+  }
+});

@omniroute/opencode-plugin/tests/multi-instance.test.ts

@@ -0,0 +1,136 @@
+/**
+ * T-08 multi-instance smoke.
+ *
+ * Validates that two `OmniRoutePlugin(input, opts)` invocations with
+ * different `providerId` values coexist without sharing mutable state.
+ * This is the contract that lets opencode.json declare prod + preprod
+ * side by side:
+ *
+ *   "plugin": [
+ *     ["@omniroute/opencode-plugin", {"providerId": "omniroute-prod",    "baseURL": "https://or.example/v1"}],
+ *     ["@omniroute/opencode-plugin", {"providerId": "omniroute-preprod", "baseURL": "https://or-preprod.example/v1"}]
+ *   ]
+ *
+ * Assertions:
+ *   - Each invocation returns its own hooks object (no identity reuse).
+ *   - Each `auth` hook carries its own `provider` matching opts.providerId.
+ *   - Each `auth.methods` array is its own array (not the same reference).
+ *   - Calling the factory twice with IDENTICAL opts still yields two
+ *     independent objects (no instance reuse / no shared closure cache).
+ *   - Mutating one instance's auth hook does NOT bleed into the other.
+ *   - Each instance's loader closure captures its OWN baseURL — no
+ *     last-write-wins module-scope state.
+ */
+
+import test from "node:test";
+import assert from "node:assert/strict";
+import { OmniRoutePlugin } from "../src/index.js";
+
+const fakeInput = {} as Parameters<typeof OmniRoutePlugin>[0];
+
+test("multi-instance: two plugin invocations bind to their own providerId", async () => {
+  const a = await OmniRoutePlugin(fakeInput, {
+    providerId: "omniroute-prod",
+    baseURL: "https://a.example/v1",
+  });
+  const b = await OmniRoutePlugin(fakeInput, {
+    providerId: "omniroute-preprod",
+    baseURL: "https://b.example/v1",
+  });
+
+  assert.equal(a.auth?.provider, "opencode-omniroute-prod");
+  assert.equal(b.auth?.provider, "opencode-omniroute-preprod");
+});
+
+test("multi-instance: hook objects + nested arrays are independent references", async () => {
+  const a = await OmniRoutePlugin(fakeInput, {
+    providerId: "alpha",
+    baseURL: "https://a.example/v1",
+  });
+  const b = await OmniRoutePlugin(fakeInput, {
+    providerId: "bravo",
+    baseURL: "https://b.example/v1",
+  });
+
+  assert.notEqual(a, b, "top-level hooks objects must not be the same reference");
+  assert.notEqual(a.auth, b.auth, "auth hooks must not be the same reference");
+  assert.notEqual(
+    a.auth?.methods,
+    b.auth?.methods,
+    "methods arrays must not be the same reference"
+  );
+});
+
+test("multi-instance: identical opts twice still yield independent objects", async () => {
+  const opts = { providerId: "twin", baseURL: "https://twin.example/v1" };
+  const first = await OmniRoutePlugin(fakeInput, { ...opts });
+  const second = await OmniRoutePlugin(fakeInput, { ...opts });
+
+  assert.notEqual(first, second);
+  assert.notEqual(first.auth, second.auth);
+  assert.notEqual(first.auth?.methods, second.auth?.methods);
+  // Same provider id is fine — what matters is no shared mutable state.
+  assert.equal(first.auth?.provider, "opencode-twin");
+  assert.equal(second.auth?.provider, "opencode-twin");
+});
+
+test("multi-instance: mutating instance A's auth.methods does not affect instance B", async () => {
+  const a = await OmniRoutePlugin(fakeInput, {
+    providerId: "iso-a",
+    baseURL: "https://a.example/v1",
+  });
+  const b = await OmniRoutePlugin(fakeInput, {
+    providerId: "iso-b",
+    baseURL: "https://b.example/v1",
+  });
+
+  const beforeLen = b.auth?.methods?.length ?? 0;
+  // Mutate a's methods array — extend it; b's must be untouched.
+  // We don't know the concrete method shape so push a sentinel cast.
+  a.auth?.methods?.push({ type: "api", label: "sentinel" } as never);
+  assert.equal(b.auth?.methods?.length, beforeLen, "instance B leaked from instance A mutation");
+});
+
+test("multi-instance: loader closures see their own opts (not last-write-wins)", async () => {
+  // Each plugin's loader builds its loader payload from the providerId/baseURL
+  // captured at invocation time. If the factory accidentally shared a closure
+  // (e.g. a module-scope let that the last invocation overwrites), both
+  // loaders would emit the same baseURL. Verify they don't.
+  const a = await OmniRoutePlugin(fakeInput, {
+    providerId: "omniroute-prod",
+    baseURL: "https://prod.example/v1",
+  });
+  const b = await OmniRoutePlugin(fakeInput, {
+    providerId: "omniroute-preprod",
+    baseURL: "https://preprod.example/v1",
+  });
+
+  assert.ok(a.auth?.loader, "instance A must have a loader");
+  assert.ok(b.auth?.loader, "instance B must have a loader");
+
+  const getAuthA = async () => ({ type: "api", key: "sk-prod" }) as never;
+  const getAuthB = async () => ({ type: "api", key: "sk-preprod" }) as never;
+
+  const rA = (await a.auth!.loader!(getAuthA, {} as never)) as Record<string, unknown>;
+  const rB = (await b.auth!.loader!(getAuthB, {} as never)) as Record<string, unknown>;
+
+  assert.equal(rA.apiKey, "sk-prod");
+  assert.equal(rA.baseURL, "https://prod.example/v1");
+  assert.equal(rB.apiKey, "sk-preprod");
+  assert.equal(rB.baseURL, "https://preprod.example/v1");
+});
+
+test("multi-instance: invalid opts on one instance does not poison the other", async () => {
+  // Sequencing: bad opts → good opts. The bad call must throw cleanly; the
+  // good call must still produce a working hooks object. Confirms no
+  // half-built module-level state survives a failed parse.
+  await assert.rejects(
+    () => OmniRoutePlugin(fakeInput, { providerId: "bad id!" } as never),
+    /providerId/
+  );
+  const ok = await OmniRoutePlugin(fakeInput, {
+    providerId: "recovered",
+    baseURL: "https://ok.example/v1",
+  });
+  assert.equal(ok.auth?.provider, "opencode-recovered");
+});

@omniroute/opencode-plugin/tests/options-schema.test.ts

@@ -0,0 +1,104 @@
+/**
+ * T-08 options-schema tests.
+ *
+ * Covers `parseOmniRoutePluginOptions(opts)` — the strict Zod gate that
+ * validates the second-arg `PluginOptions` bag from opencode.json before
+ * any hook is wired. Anti-pattern checklist mirrored here:
+ *
+ *  - `null` / `undefined` must collapse to `{}` (defaults apply downstream).
+ *  - Unknown keys must THROW (`.strict()` catches opencode.json typos).
+ *  - Validation runs at parse time, not import time (module loads cleanly).
+ */
+
+import test from "node:test";
+import assert from "node:assert/strict";
+import { parseOmniRoutePluginOptions } from "../src/index.js";
+
+test("parseOmniRoutePluginOptions: undefined → {}", () => {
+  assert.deepEqual(parseOmniRoutePluginOptions(undefined), {});
+});
+
+test("parseOmniRoutePluginOptions: null → {}", () => {
+  assert.deepEqual(parseOmniRoutePluginOptions(null), {});
+});
+
+test("parseOmniRoutePluginOptions: empty object → {}", () => {
+  assert.deepEqual(parseOmniRoutePluginOptions({}), {});
+});
+
+test("parseOmniRoutePluginOptions: valid providerId → returns it", () => {
+  const r = parseOmniRoutePluginOptions({ providerId: "omniroute-preprod" });
+  assert.equal(r.providerId, "omniroute-preprod");
+});
+
+test("parseOmniRoutePluginOptions: invalid providerId (special chars) → throws", () => {
+  assert.throws(
+    () => parseOmniRoutePluginOptions({ providerId: "omniroute prod!" }),
+    /providerId.*slug/i
+  );
+});
+
+test("parseOmniRoutePluginOptions: empty providerId → throws", () => {
+  assert.throws(() => parseOmniRoutePluginOptions({ providerId: "" }), /providerId/i);
+});
+
+test("parseOmniRoutePluginOptions: valid modelCacheTtl → returns it", () => {
+  const r = parseOmniRoutePluginOptions({ modelCacheTtl: 60_000 });
+  assert.equal(r.modelCacheTtl, 60_000);
+});
+
+test("parseOmniRoutePluginOptions: negative modelCacheTtl → throws", () => {
+  assert.throws(() => parseOmniRoutePluginOptions({ modelCacheTtl: -1 }), /modelCacheTtl/i);
+});
+
+test("parseOmniRoutePluginOptions: zero modelCacheTtl → throws (positive required)", () => {
+  assert.throws(() => parseOmniRoutePluginOptions({ modelCacheTtl: 0 }), /modelCacheTtl/i);
+});
+
+test("parseOmniRoutePluginOptions: invalid baseURL (not a URL) → throws", () => {
+  assert.throws(() => parseOmniRoutePluginOptions({ baseURL: "not-a-url" }), /baseURL/i);
+});
+
+test("parseOmniRoutePluginOptions: unknown key → throws (strict mode catches typos)", () => {
+  assert.throws(
+    () =>
+      parseOmniRoutePluginOptions({
+        providerId: "omniroute",
+        provider_id: "typo-here",
+      }),
+    /provider_id|unrecognized/i
+  );
+});
+
+test("parseOmniRoutePluginOptions: all four fields populated correctly → returns them", () => {
+  const opts = {
+    providerId: "omniroute-prod",
+    displayName: "OmniRoute Production",
+    modelCacheTtl: 120_000,
+    baseURL: "https://or.example.com/v1",
+  };
+  const r = parseOmniRoutePluginOptions(opts);
+  assert.deepEqual(r, opts);
+});
+
+test("parseOmniRoutePluginOptions: error message lists every issue path", () => {
+  // Two bad fields at once → error string should mention BOTH.
+  try {
+    parseOmniRoutePluginOptions({
+      providerId: "",
+      baseURL: "garbage",
+    });
+    assert.fail("expected throw");
+  } catch (err) {
+    const msg = (err as Error).message;
+    assert.match(msg, /providerId/);
+    assert.match(msg, /baseURL/);
+  }
+});
+
+test("parseOmniRoutePluginOptions: module import alone does NOT throw", async () => {
+  // Re-importing the entry must not trigger validation; validation only fires
+  // on explicit parseOmniRoutePluginOptions / OmniRoutePlugin invocation.
+  const mod = await import("../src/index.js");
+  assert.equal(typeof mod.parseOmniRoutePluginOptions, "function");
+});

@omniroute/opencode-plugin/tests/provider.test.ts

@@ -0,0 +1,271 @@
+/**
+ * T-03 provider-hook contract tests.
+ *
+ * Covers `createOmniRouteProviderHook(opts, deps)`:
+ *   - hook.id binds to resolved providerId (single + multi-instance)
+ *   - models() narrows ctx.auth, fetches via injected fetcher, caches per
+ *     (baseURL, apiKey) tuple, refetches after TTL
+ *   - mapRawModelToModelV2 emits a v2 Model shape matching the
+ *     @opencode-ai/sdk/v2 type
+ *
+ * Mocking strategy: the fetcher is dependency-injected at hook construction
+ * (`deps.fetcher`). No global fetch monkey-patch needed. `deps.now` lets us
+ * fast-forward time deterministically for TTL assertions.
+ */
+
+import test from "node:test";
+import assert from "node:assert/strict";
+import {
+  createOmniRouteProviderHook,
+  mapRawModelToModelV2,
+  type OmniRouteRawModelEntry,
+  type OmniRouteModelsFetcher,
+} from "../src/index.js";
+
+const FIXTURE: OmniRouteRawModelEntry[] = [
+  {
+    id: "claude-primary",
+    object: "model",
+    owned_by: "combo",
+    capabilities: { tool_calling: true, reasoning: true, vision: true, thinking: true },
+    context_length: 200000,
+    max_output_tokens: 64000,
+    input_modalities: ["text", "image"],
+    output_modalities: ["text"],
+  },
+  {
+    id: "claude-low",
+    object: "model",
+    owned_by: "combo",
+    capabilities: { tool_calling: true, reasoning: true, vision: true, thinking: false },
+    context_length: 200000,
+    max_output_tokens: 64000,
+    input_modalities: ["text", "image"],
+    output_modalities: ["text"],
+  },
+  {
+    id: "gemini-3-flash",
+    object: "model",
+    owned_by: "google",
+    capabilities: { tool_calling: true, reasoning: false, vision: true, thinking: false },
+    context_length: 1000000,
+    max_output_tokens: 8192,
+    input_modalities: ["text", "image"],
+    output_modalities: ["text"],
+  },
+];
+
+function stubFetcher(payload: OmniRouteRawModelEntry[]): OmniRouteModelsFetcher & {
+  callCount: () => number;
+  callsBy: () => Array<[string, string]>;
+} {
+  let calls: Array<[string, string]> = [];
+  const f: OmniRouteModelsFetcher = async (baseURL, apiKey) => {
+    calls.push([baseURL, apiKey]);
+    return payload;
+  };
+  return Object.assign(f, {
+    callCount: () => calls.length,
+    callsBy: () => calls,
+  });
+}
+
+const apiAuth = (key: string, baseURL?: string): unknown =>
+  baseURL ? { type: "api", key, baseURL } : { type: "api", key };
+
+test("createOmniRouteProviderHook: default providerId is 'omniroute'", () => {
+  const hook = createOmniRouteProviderHook(undefined, { combosFetcher: async () => [] });
+  assert.equal(hook.id, "opencode-omniroute");
+});
+
+test("createOmniRouteProviderHook: custom providerId binds to hook.id (multi-instance)", () => {
+  const a = createOmniRouteProviderHook(
+    { providerId: "omniroute-preprod" },
+    { combosFetcher: async () => [] }
+  );
+  const b = createOmniRouteProviderHook(
+    { providerId: "omniroute-local" },
+    { combosFetcher: async () => [] }
+  );
+  assert.equal(a.id, "opencode-omniroute-preprod");
+  assert.equal(b.id, "opencode-omniroute-local");
+});
+
+test("models: extracts apiKey from ctx.auth (type=api) and calls fetcher with it", async () => {
+  const fetcher = stubFetcher(FIXTURE);
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1" },
+    { fetcher, combosFetcher: async () => [] }
+  );
+  const out = await hook.models!({} as never, { auth: apiAuth("sk-abc") as never });
+  assert.equal(fetcher.callCount(), 1);
+  assert.deepEqual(fetcher.callsBy()[0], ["https://or.example.com/v1", "sk-abc"]);
+  assert.equal(Object.keys(out).length, 3);
+  assert.ok(out["opencode-omniroute/claude-primary"]);
+});
+
+test("models: returns {} when ctx.auth is null/undefined/wrong-type/empty-key", async () => {
+  const fetcher = stubFetcher(FIXTURE);
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1" },
+    { fetcher, combosFetcher: async () => [] }
+  );
+
+  assert.deepEqual(await hook.models!({} as never, {} as never), {});
+  assert.deepEqual(await hook.models!({} as never, { auth: undefined } as never), {});
+  assert.deepEqual(
+    await hook.models!({} as never, {
+      auth: { type: "oauth", refresh: "r", access: "a", expires: 0 } as never,
+    }),
+    {}
+  );
+  assert.deepEqual(
+    await hook.models!({} as never, { auth: { type: "api", key: "" } as never }),
+    {}
+  );
+  assert.equal(fetcher.callCount(), 0, "fetcher must not be called on auth rejection");
+});
+
+test("models: returns {} when no baseURL resolvable (no opts.baseURL and no auth.baseURL)", async () => {
+  const fetcher = stubFetcher(FIXTURE);
+  const hook = createOmniRouteProviderHook({}, { fetcher, combosFetcher: async () => [] });
+  // valid api auth but neither opts nor auth carries a baseURL
+  assert.deepEqual(await hook.models!({} as never, { auth: apiAuth("sk-x") as never }), {});
+  assert.equal(fetcher.callCount(), 0);
+});
+
+test("models: baseURL falls back to auth.baseURL when opts.baseURL absent", async () => {
+  const fetcher = stubFetcher(FIXTURE);
+  const hook = createOmniRouteProviderHook({}, { fetcher, combosFetcher: async () => [] });
+  const out = await hook.models!({} as never, {
+    auth: apiAuth("sk-y", "https://or.creds-attached.example/v1") as never,
+  });
+  assert.equal(fetcher.callCount(), 1);
+  assert.equal(fetcher.callsBy()[0][0], "https://or.creds-attached.example/v1");
+  assert.equal(Object.keys(out).length, 3);
+});
+
+test("models: maps a sample /v1/models entry to ModelV2 (sanity)", async () => {
+  const fetcher = stubFetcher(FIXTURE);
+  const hook = createOmniRouteProviderHook(
+    { providerId: "omniroute", baseURL: "https://or.example.com/v1" },
+    { fetcher, combosFetcher: async () => [] }
+  );
+  const out = await hook.models!({} as never, { auth: apiAuth("sk-abc") as never });
+  const claude = out["opencode-omniroute/claude-primary"];
+  assert.ok(claude, "claude-primary present");
+  // `mapRawModelToModelV2` stamps the provider prefix on the id so OC's
+  // static-catalog reader resolves `(providerID, modelID)` from the key.
+  assert.equal(claude.id, "opencode-omniroute/claude-primary");
+  assert.equal(claude.name, "claude-primary");
+  assert.equal(claude.providerID, "opencode-omniroute");
+  assert.equal(claude.api.id, "openai-compatible");
+  assert.equal(claude.api.url, "https://or.example.com/v1");
+  assert.equal(claude.api.npm, "@ai-sdk/openai-compatible");
+  // capabilities: toolcall (one word), reasoning OR thinking, attachment = vision
+  assert.equal(claude.capabilities.toolcall, true);
+  assert.equal(claude.capabilities.reasoning, true);
+  assert.equal(claude.capabilities.attachment, true);
+  assert.equal(claude.capabilities.temperature, true);
+  // modalities mapped from arrays
+  assert.equal(claude.capabilities.input.text, true);
+  assert.equal(claude.capabilities.input.image, true);
+  assert.equal(claude.capabilities.input.audio, false);
+  assert.equal(claude.capabilities.output.text, true);
+  assert.equal(claude.capabilities.output.image, false);
+  // cost is zeroed (OmniRoute /v1/models has no pricing)
+  assert.deepEqual(claude.cost, { input: 0, output: 0, cache: { read: 0, write: 0 } });
+  // limits
+  assert.equal(claude.limit.context, 200000);
+  assert.equal(claude.limit.output, 64000);
+  assert.equal(claude.status, "active");
+});
+
+test("mapRawModelToModelV2: thinking-only model still surfaces reasoning=true", () => {
+  const m = mapRawModelToModelV2(
+    {
+      id: "thinking-only",
+      capabilities: { thinking: true, reasoning: false },
+      context_length: 100000,
+      max_output_tokens: 8192,
+    },
+    { providerId: "omniroute", baseURL: "https://or.example.com/v1" }
+  );
+  assert.equal(m.capabilities.reasoning, true);
+});
+
+test("mapRawModelToModelV2: missing capabilities defaults to all-false (except temperature)", () => {
+  const m = mapRawModelToModelV2(
+    { id: "minimal" },
+    { providerId: "omniroute", baseURL: "https://or.example.com/v1" }
+  );
+  assert.equal(m.capabilities.temperature, true);
+  assert.equal(m.capabilities.reasoning, false);
+  assert.equal(m.capabilities.attachment, false);
+  assert.equal(m.capabilities.toolcall, false);
+  // default modalities = text only
+  assert.equal(m.capabilities.input.text, true);
+  assert.equal(m.capabilities.output.text, true);
+  // missing context / output tokens → 0 fallback (ModelV2.limit.{context,output} required)
+  assert.equal(m.limit.context, 0);
+  assert.equal(m.limit.output, 0);
+});
+
+test("models: caches result for second call within TTL (fetcher called once)", async () => {
+  const fetcher = stubFetcher(FIXTURE);
+  let nowMs = 1_000_000;
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1", modelCacheTtl: 60_000 },
+    { fetcher, now: () => nowMs, combosFetcher: async () => [] }
+  );
+
+  const a = await hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  nowMs += 30_000; // half the TTL
+  const b = await hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  assert.equal(fetcher.callCount(), 1, "second call within TTL must hit the cache");
+  assert.equal(Object.keys(a).length, 3);
+  assert.equal(Object.keys(b).length, 3);
+});
+
+test("models: refetches after TTL expires", async () => {
+  const fetcher = stubFetcher(FIXTURE);
+  let nowMs = 1_000_000;
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1", modelCacheTtl: 60_000 },
+    { fetcher, now: () => nowMs, combosFetcher: async () => [] }
+  );
+
+  await hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  nowMs += 60_001; // just past the TTL
+  await hook.models!({} as never, { auth: apiAuth("sk-z") as never });
+  assert.equal(fetcher.callCount(), 2, "call past TTL must refetch");
+});
+
+test("models: caches per (baseURL, apiKey) tuple (different keys → independent fetches)", async () => {
+  const fetcher = stubFetcher(FIXTURE);
+  const hook = createOmniRouteProviderHook(
+    { baseURL: "https://or.example.com/v1", modelCacheTtl: 300_000 },
+    { fetcher, combosFetcher: async () => [] }
+  );
+
+  await hook.models!({} as never, { auth: apiAuth("sk-A") as never });
+  await hook.models!({} as never, { auth: apiAuth("sk-B") as never });
+  await hook.models!({} as never, { auth: apiAuth("sk-A") as never }); // cached
+  await hook.models!({} as never, { auth: apiAuth("sk-B") as never }); // cached
+  assert.equal(fetcher.callCount(), 2, "one fetch per distinct apiKey, then cache hits");
+});
+
+test("models: caches per (baseURL, apiKey) tuple (different baseURL → independent fetches)", async () => {
+  const fetcher = stubFetcher(FIXTURE);
+  const hook = createOmniRouteProviderHook(
+    { modelCacheTtl: 300_000 }, // no opts.baseURL → falls back to auth.baseURL
+    { fetcher, combosFetcher: async () => [] }
+  );
+
+  await hook.models!({} as never, { auth: apiAuth("sk-same", "https://prod.example/v1") as never });
+  await hook.models!({} as never, {
+    auth: apiAuth("sk-same", "https://preprod.example/v1") as never,
+  });
+  await hook.models!({} as never, { auth: apiAuth("sk-same", "https://prod.example/v1") as never }); // cached
+  assert.equal(fetcher.callCount(), 2, "distinct baseURLs share apiKey but not cache");
+});

@omniroute/opencode-plugin/tests/scaffold.test.ts

@@ -0,0 +1,73 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+import {
+  OmniRoutePlugin,
+  OMNIROUTE_PROVIDER_KEY,
+  DEFAULT_MODEL_CACHE_TTL_MS,
+  resolveOmniRoutePluginOptions,
+} from "../src/index.js";
+
+test("scaffold: exports public surface", () => {
+  assert.equal(
+    typeof OmniRoutePlugin,
+    "function",
+    "OmniRoutePlugin must be a function (Plugin factory)"
+  );
+  assert.equal(OMNIROUTE_PROVIDER_KEY, "omniroute");
+  assert.equal(DEFAULT_MODEL_CACHE_TTL_MS, 300_000);
+});
+
+test("scaffold: default export is v1 plugin shape { id, server: OmniRoutePlugin }", async () => {
+  const mod = await import("../src/index.js");
+  assert.equal(typeof mod.default, "object");
+  assert.equal(mod.default.id, "@omniroute/opencode-plugin");
+  assert.equal(mod.default.server, mod.OmniRoutePlugin);
+});
+
+test("resolveOmniRoutePluginOptions: defaults", () => {
+  const r = resolveOmniRoutePluginOptions();
+  assert.equal(r.providerId, "opencode-omniroute");
+  assert.equal(r.displayName, "OmniRoute");
+  assert.equal(r.modelCacheTtl, 300_000);
+  assert.equal(r.baseURL, undefined);
+});
+
+test("resolveOmniRoutePluginOptions: custom providerId derives displayName", () => {
+  const r = resolveOmniRoutePluginOptions({ providerId: "omniroute-preprod" });
+  assert.equal(r.providerId, "opencode-omniroute-preprod");
+  assert.equal(r.displayName, "OmniRoute (opencode-omniroute-preprod)");
+});
+
+test("resolveOmniRoutePluginOptions: explicit displayName wins", () => {
+  const r = resolveOmniRoutePluginOptions({
+    providerId: "omniroute-x",
+    displayName: "Custom Label",
+  });
+  assert.equal(r.displayName, "Custom Label");
+});
+
+test("resolveOmniRoutePluginOptions: invalid TTL falls back to default", () => {
+  assert.equal(resolveOmniRoutePluginOptions({ modelCacheTtl: 0 }).modelCacheTtl, 300_000);
+  assert.equal(resolveOmniRoutePluginOptions({ modelCacheTtl: -1 }).modelCacheTtl, 300_000);
+});
+
+test("resolveOmniRoutePluginOptions: positive TTL respected", () => {
+  assert.equal(resolveOmniRoutePluginOptions({ modelCacheTtl: 60_000 }).modelCacheTtl, 60_000);
+});
+
+test("OmniRoutePlugin: returns an empty hooks object (scaffold)", async () => {
+  const fakeCtx = {} as Parameters<typeof OmniRoutePlugin>[0];
+  const hooks = await OmniRoutePlugin(fakeCtx);
+  assert.equal(typeof hooks, "object");
+  assert.notEqual(hooks, null);
+});
+
+test("scaffold: built ESM default export resolves with the v1 plugin shape", async () => {
+  // The plugin is ESM-only now — the CJS bundle was dropped to fix the OpenCode
+  // loader (#3883), so there is no more ../dist/index.cjs. Validate that the built
+  // distributable's default export still carries the OpenCode v1 { id, server } shape.
+  const mod = await import("../dist/index.js");
+  assert.strictEqual(typeof mod.default, "object");
+  assert.strictEqual(mod.default.id, "@omniroute/opencode-plugin");
+  assert.strictEqual(typeof mod.default.server, "function");
+});

@omniroute/opencode-plugin/tests/usable-combo.test.ts

@@ -0,0 +1,85 @@
+/**
+ * Regression tests for `isUsableCombo` (release/v3.8.2 code review, finding C1).
+ *
+ * The combo member refs returned by `/api/combos` do NOT carry a separate
+ * `providerId` field — OmniRoute's `normalizeComboRecord` folds the provider
+ * id INTO the full model string (e.g. "cc/claude-opus-4-7"). The previous
+ * implementation read `step.providerId` (always `undefined`), so the
+ * `usableOnly` combo filter silently never dropped anything. These tests pin
+ * the corrected behavior: the verdict is derived from the `step.model` prefix,
+ * mirroring `isUsableRawModelId`'s subtract-filter semantics.
+ */
+
+import test from "node:test";
+import assert from "node:assert/strict";
+
+import { isUsableCombo, type OmniRouteRawCombo } from "../src/index.js";
+
+/** Build a `usable` set bundle for the tests. */
+function buildUsable(opts: { aliases?: string[]; canonicals?: string[]; known?: string[] }): {
+  aliases: Set<string>;
+  canonicals: Set<string>;
+  knownAliases: Set<string>;
+} {
+  return {
+    aliases: new Set(opts.aliases ?? []),
+    canonicals: new Set(opts.canonicals ?? []),
+    // knownAliases is the union of every prefix the universe is aware of —
+    // usable or not. Default to including the usable aliases too.
+    knownAliases: new Set([...(opts.known ?? []), ...(opts.aliases ?? [])]),
+  };
+}
+
+function combo(models: OmniRouteRawCombo["models"]): OmniRouteRawCombo {
+  return { id: "c1", name: "Test Combo", models };
+}
+
+test("isUsableCombo: member with a usable alias prefix → keep", () => {
+  const usable = buildUsable({ aliases: ["cc"], known: ["cc", "dead"] });
+  const c = combo([{ kind: "model", model: "cc/claude-opus-4-7" }]);
+  assert.equal(isUsableCombo(c, usable), true);
+});
+
+test("isUsableCombo: all members known-but-NOT-usable → drop (the C1 regression)", () => {
+  // Before the fix this returned true unconditionally because step.providerId
+  // was always undefined. Now the known-but-unusable "dead" prefix is dropped.
+  const usable = buildUsable({ aliases: ["cc"], known: ["cc", "dead"] });
+  const c = combo([
+    { kind: "model", model: "dead/legacy-model" },
+    { kind: "model", model: "dead/another" },
+  ]);
+  assert.equal(isUsableCombo(c, usable), false);
+});
+
+test("isUsableCombo: unknown prefix → keep (cannot prove unroutable)", () => {
+  const usable = buildUsable({ aliases: ["cc"], known: ["cc", "dead"] });
+  const c = combo([{ kind: "model", model: "agentrouter/mystery" }]);
+  assert.equal(isUsableCombo(c, usable), true);
+});
+
+test("isUsableCombo: mixed non-usable + usable member → keep", () => {
+  const usable = buildUsable({ aliases: ["cc"], known: ["cc", "dead"] });
+  const c = combo([
+    { kind: "model", model: "dead/legacy" },
+    { kind: "model", model: "cc/claude-opus-4-7" },
+  ]);
+  assert.equal(isUsableCombo(c, usable), true);
+});
+
+test("isUsableCombo: zero members → keep", () => {
+  const usable = buildUsable({ aliases: ["cc"], known: ["cc"] });
+  assert.equal(isUsableCombo(combo([]), usable), true);
+  assert.equal(isUsableCombo(combo(undefined), usable), true);
+});
+
+test("isUsableCombo: only combo-ref steps (no resolvable model) → keep", () => {
+  const usable = buildUsable({ aliases: ["cc"], known: ["cc", "dead"] });
+  const c = combo([{ kind: "combo-ref", comboName: "nested" }]);
+  assert.equal(isUsableCombo(c, usable), true);
+});
+
+test("isUsableCombo: usable canonical prefix → keep", () => {
+  const usable = buildUsable({ canonicals: ["anthropic"], known: ["anthropic", "dead"] });
+  const c = combo([{ kind: "model", model: "anthropic/claude-opus-4-7" }]);
+  assert.equal(isUsableCombo(c, usable), true);
+});

@omniroute/opencode-plugin/tsconfig.json

@@ -0,0 +1,20 @@
+{
+  "compilerOptions": {
+    "target": "ES2022",
+    "module": "ESNext",
+    "moduleResolution": "Bundler",
+    "lib": ["ES2022"],
+    "types": ["node"],
+    "strict": true,
+    "esModuleInterop": true,
+    "skipLibCheck": true,
+    "declaration": true,
+    "isolatedModules": true,
+    "forceConsistentCasingInFileNames": true,
+    "noUncheckedIndexedAccess": false,
+    "outDir": "dist",
+    "rootDir": "src"
+  },
+  "include": ["src/**/*.ts"],
+  "exclude": ["dist", "node_modules", "tests"]
+}

@omniroute/opencode-plugin/tsup.config.ts

@@ -0,0 +1,20 @@
+import { defineConfig } from "tsup";
+
+export default defineConfig({
+  entry: ["src/index.ts"],
+  format: ["esm"],
+  dts: true,
+  clean: true,
+  sourcemap: false,
+  splitting: false,
+  treeshake: false,
+  target: "node22",
+  outDir: "dist",
+  minify: false,
+  cjsInterop: false,
+  // Bundle runtime deps so the .tgz / npm install is self-contained.
+  // `zod` is required at runtime by the options schema and would otherwise
+  // need a peer install when the plugin is loaded directly from a file path
+  // in opencode.jsonc.
+  noExternal: ["zod"],
+});

@omniroute/opencode-provider/.gitignore

@@ -0,0 +1,4 @@
+node_modules
+dist
+*.log
+.DS_Store

@omniroute/opencode-provider/.npmignore

@@ -0,0 +1,7 @@
+src
+tests
+tsconfig.json
+tsup.config.ts
+node_modules
+.DS_Store
+*.log

@omniroute/opencode-provider/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 OmniRoute contributors
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

@omniroute/opencode-provider/README.md

@@ -0,0 +1,156 @@
+# @omniroute/opencode-provider
+
+> ## ⚠️ Deprecated — use [`@omniroute/opencode-plugin`](https://www.npmjs.com/package/@omniroute/opencode-plugin) instead
+>
+> This package writes a **static** `provider.omniroute` block to `opencode.json` from a hardcoded default model list, so it **drifts behind your live OmniRoute catalog** — adding a model in OmniRoute won't show up in OpenCode until you re-run the generator, and OpenCode Desktop/Web only surfaces a subset of the static models.
+>
+> **`@omniroute/opencode-plugin`** solves this by fetching `GET /v1/models` from your OmniRoute instance at OpenCode startup, so the model list is always live (see [#3419](https://github.com/diegosouzapw/OmniRoute/issues/3419)). It is now the recommended path.
+>
+> **One-line migration** — replace the static `provider.omniroute` block in `opencode.json` with a single plugin entry:
+>
+> ```jsonc
+> // opencode.json
+> {
+>   "$schema": "https://opencode.ai/config.json",
+>   "plugin": ["@omniroute/opencode-plugin"]
+> }
+> ```
+>
+> This package is **not removed** and still works for static/offline config generation, but it is no longer actively recommended and won't track new models automatically.
+
+Helper for connecting [OpenCode](https://opencode.ai) to a running [OmniRoute](https://github.com/diegosouzapw/OmniRoute) AI gateway.
+
+The package emits a **schema-valid entry** for `opencode.json` (`https://opencode.ai/config.json`) that delegates the actual runtime to [`@ai-sdk/openai-compatible`](https://www.npmjs.com/package/@ai-sdk/openai-compatible). It does not ship any new HTTP client — OmniRoute already exposes an OpenAI-compatible surface, and OpenCode already speaks it through the AI SDK.
+
+> Pre-1.0. The API may still change. See `CHANGELOG` in the OmniRoute repo for breaking notes.
+
+## Installation
+
+```bash
+npm install --save-dev @omniroute/opencode-provider
+# or
+pnpm add -D @omniroute/opencode-provider
+```
+
+You also need OpenCode's own runtime dep, but that's a transitive concern — OpenCode itself ships with `@ai-sdk/openai-compatible`. This package only **generates configuration**.
+
+## Quick start
+
+### 1. Scaffold a fresh `opencode.json`
+
+```ts
+import { writeFileSync } from "node:fs";
+import { buildOmniRouteOpenCodeConfig } from "@omniroute/opencode-provider";
+
+const config = buildOmniRouteOpenCodeConfig({
+  baseURL: "http://localhost:20128", // or your OmniRoute deployment URL
+  apiKey: process.env.OMNIROUTE_API_KEY ?? "sk_omniroute",
+});
+
+writeFileSync("opencode.json", JSON.stringify(config, null, 2));
+```
+
+The resulting `opencode.json`:
+
+```jsonc
+{
+  "$schema": "https://opencode.ai/config.json",
+  "provider": {
+    "omniroute": {
+      "npm": "@ai-sdk/openai-compatible",
+      "name": "OmniRoute",
+      "options": {
+        "baseURL": "http://localhost:20128/v1",
+        "apiKey": "sk_omniroute",
+      },
+      "models": {
+        "claude-opus-4-5-thinking": { "name": "claude-opus-4-5-thinking" },
+        "claude-sonnet-4-5-thinking": { "name": "claude-sonnet-4-5-thinking" },
+        "gemini-3.1-pro-high": { "name": "gemini-3.1-pro-high" },
+        "gemini-3-flash": { "name": "gemini-3-flash" },
+      },
+    },
+  },
+}
+```
+
+### 2. Merge into an existing `opencode.json`
+
+```ts
+import { createOmniRouteProvider } from "@omniroute/opencode-provider";
+
+const provider = createOmniRouteProvider({
+  baseURL: "http://localhost:20128",
+  apiKey: process.env.OMNIROUTE_API_KEY!,
+});
+
+// Place `provider` under provider.omniroute in your opencode.json
+```
+
+If you already have an `opencode.json` on disk and want a non-destructive merge from the OmniRoute side, use `omniroute config opencode` from the CLI (ships with the main OmniRoute install) — it preserves comments and unrelated keys.
+
+## API
+
+### `createOmniRouteProvider(options): OpenCodeProviderEntry`
+
+Returns the value to place under `provider.omniroute` inside `opencode.json`.
+
+| Option        | Type                    | Required | Description                                                                                                  |
+| ------------- | ----------------------- | -------- | ------------------------------------------------------------------------------------------------------------ |
+| `baseURL`     | `string`                | Yes      | OmniRoute base URL. Accepts `http://host:port` **or** `http://host:port/v1`. Trailing slashes are tolerated. |
+| `apiKey`      | `string`                | Yes      | OmniRoute API key. Use `sk_omniroute` for local installs that have `REQUIRE_API_KEY=false`.                  |
+| `displayName` | `string`                | No       | Custom name shown in the OpenCode UI. Default: `"OmniRoute"`.                                                |
+| `models`      | `string[]`              | No       | Override the surfaced model catalog. Default: 4 curated models — see `OMNIROUTE_DEFAULT_OPENCODE_MODELS`.    |
+| `modelLabels` | `Record<string,string>` | No       | Human-readable labels keyed by model id.                                                                     |
+
+Throws on empty/invalid input — `baseURL` must be a real URL, `apiKey` must be a non-empty string.
+
+### `buildOmniRouteOpenCodeConfig(options): OpenCodeConfigDocument`
+
+Same options as above, but returns a full document with `$schema` and the `provider.omniroute` wrapper, ready to write to `opencode.json`.
+
+### `normalizeBaseURL(input): string`
+
+Exported for completeness. Strips trailing `/`, deduplicates a trailing `/v1`, and re-appends exactly one `/v1`. Throws on empty / non-URL input.
+
+### Constants
+
+- `OMNIROUTE_PROVIDER_KEY` — `"omniroute"` (the key used under `provider.*`).
+- `OMNIROUTE_PROVIDER_NPM` — `"@ai-sdk/openai-compatible"` (the runtime delegate).
+- `OPENCODE_CONFIG_SCHEMA` — `"https://opencode.ai/config.json"`.
+- `OMNIROUTE_DEFAULT_OPENCODE_MODELS` — readonly list of default model ids.
+
+## Custom model catalog
+
+```ts
+import { createOmniRouteProvider } from "@omniroute/opencode-provider";
+
+createOmniRouteProvider({
+  baseURL: "http://localhost:20128",
+  apiKey: "sk_omniroute",
+  models: ["auto", "claude-opus-4-8", "gpt-5.5"],
+  modelLabels: {
+    auto: "Auto-Combo (recommended)",
+    "claude-opus-4-8": "Claude Opus 4.8",
+    "gpt-5.5": "GPT-5.5",
+  },
+});
+```
+
+Duplicates and empty strings are dropped automatically, and order is preserved.
+
+## Troubleshooting
+
+- **Requests 404 with `/v1/v1/...`** — you're on an old version (≤1.0.0). Update to `≥0.1.0` of this re-released package. The new build normalises `baseURL` automatically.
+- **`401 Invalid API key`** — your OmniRoute instance has `REQUIRE_API_KEY=true` but the key you supplied doesn't exist there. Create one via the dashboard or set `REQUIRE_API_KEY=false` and use `sk_omniroute`.
+- **OpenCode complains the provider has no models** — supply an explicit `models` list; the default 4 may be hidden by your provider visibility settings.
+
+## Related
+
+- [OmniRoute](https://github.com/diegosouzapw/OmniRoute) — the AI gateway this plugin targets.
+- [OpenCode](https://opencode.ai) — the agentic CLI consumer.
+- [`@ai-sdk/openai-compatible`](https://www.npmjs.com/package/@ai-sdk/openai-compatible) — the runtime delegate that actually speaks HTTP.
+
+## License
+
+MIT — see [`LICENSE`](./LICENSE).

@omniroute/opencode-provider/package.json

@@ -0,0 +1,63 @@
+{
+  "name": "@omniroute/opencode-provider",
+  "version": "0.1.0",
+  "description": "DEPRECATED — use @omniroute/opencode-plugin instead (it fetches the live OmniRoute /v1/models catalog at startup, so models never drift). This static-config generator still works but is no longer the recommended path. OpenCode provider helper for the OmniRoute AI Gateway.",
+  "type": "module",
+  "main": "./dist/index.cjs",
+  "module": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js",
+      "require": "./dist/index.cjs"
+    }
+  },
+  "files": [
+    "dist",
+    "README.md",
+    "LICENSE"
+  ],
+  "scripts": {
+    "build": "tsup",
+    "clean": "rm -rf dist",
+    "test": "node --import tsx/esm --test tests/index.test.ts",
+    "prepublishOnly": "npm run clean && npm run build && npm test"
+  },
+  "keywords": [
+    "omniroute",
+    "opencode",
+    "opencode-ai",
+    "ai-sdk",
+    "openai-compatible",
+    "provider",
+    "ai-gateway",
+    "llm-router"
+  ],
+  "author": "OmniRoute contributors",
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/diegosouzapw/OmniRoute.git",
+    "directory": "@omniroute/opencode-provider"
+  },
+  "bugs": {
+    "url": "https://github.com/diegosouzapw/OmniRoute/issues"
+  },
+  "homepage": "https://github.com/diegosouzapw/OmniRoute/tree/main/%40omniroute/opencode-provider#readme",
+  "engines": {
+    "node": ">=22.22.3"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "devDependencies": {
+    "@types/node": "^22.19.19",
+    "tsup": "^8.5.1",
+    "tsx": "^4.22.3",
+    "typescript": "^5.9.3"
+  },
+  "overrides": {
+    "esbuild": "^0.28.1"
+  }
+}

@omniroute/opencode-provider/src/index.ts

@@ -0,0 +1,908 @@
+/**
+ * OpenCode provider plugin for OmniRoute AI Gateway.
+ *
+ * Generates an OpenCode-compatible provider object that points to a running
+ * OmniRoute instance. The output follows the OpenCode config schema
+ * (https://opencode.ai/config.json) and delegates the runtime to
+ * `@ai-sdk/openai-compatible` so OpenCode can drive any OmniRoute-exposed
+ * model through its standard OpenAI-compatible client.
+ *
+ * Two ways to consume the helper:
+ *
+ *  1. As code, when you build your own opencode.json programmatically:
+ *
+ *     ```ts
+ *     import { buildOmniRouteOpenCodeConfig } from "@omniroute/opencode-provider";
+ *     const config = buildOmniRouteOpenCodeConfig({
+ *       baseURL: "http://localhost:20128",
+ *       apiKey: "sk_omniroute",
+ *     });
+ *     // config -> { $schema, provider: { omniroute: { npm, name, options, models } } }
+ *     ```
+ *
+ *  2. As a single-provider entry to merge into an existing opencode.json:
+ *
+ *     ```ts
+ *     import { createOmniRouteProvider } from "@omniroute/opencode-provider";
+ *     const provider = createOmniRouteProvider({ baseURL, apiKey });
+ *     // provider -> the value to place under provider.omniroute in opencode.json
+ *     ```
+ *
+ * Note: `baseURL` accepts both `http://host:port` and `http://host:port/v1`.
+ * The helper normalises trailing slashes / `/v1` so you never get `/v1/v1`.
+ */
+
+export const OMNIROUTE_PROVIDER_KEY = "omniroute" as const;
+export const OMNIROUTE_PROVIDER_NPM = "@ai-sdk/openai-compatible" as const;
+export const OPENCODE_CONFIG_SCHEMA = "https://opencode.ai/config.json" as const;
+
+/**
+ * Default catalog of models surfaced to OpenCode when the caller does not
+ * supply an explicit `models` list.
+ *
+ * Curated set covering the most commonly deployed OmniRoute models. Synced
+ * with the Alph4d0g/opencode-omniroute-auth OMNIROUTE_DEFAULT_MODELS constant
+ * (https://github.com/Alph4d0g/opencode-omniroute-auth, MIT) and extended
+ * with Claude Code passthrough models (`cc/` prefix).
+ */
+export const OMNIROUTE_DEFAULT_OPENCODE_MODELS = [
+  "cc/claude-opus-4-8",
+  "cc/claude-opus-4-7",
+  "cc/claude-sonnet-4-6",
+  "cc/claude-haiku-4-5-20251001",
+  "claude-opus-4-5-thinking",
+  "claude-sonnet-4-5-thinking",
+  "gemini-3.1-pro-high",
+  "gemini-3-flash",
+] as const;
+
+/**
+ * Optional capability flags surfaced to OpenCode's model picker.
+ *
+ * OpenCode reads these per-model keys (snake_case in JSON) to render badges
+ * and to gate features such as image attachments, reasoning mode, temperature
+ * controls and tool-calling. Omitted flags default to OpenCode's heuristics.
+ *
+ * Mirrors the capability shape used by Alph4d0g/opencode-omniroute-auth
+ * (https://github.com/Alph4d0g/opencode-omniroute-auth, MIT).
+ */
+export interface ModelCapabilities {
+  /** Display label shown in the model picker. Falls back to the model id. */
+  label?: string;
+  /** Model accepts image / file attachments. */
+  attachment?: boolean;
+  /** Model exposes a "reasoning" / extended-thinking surface. */
+  reasoning?: boolean;
+  /** Model honours the `temperature` parameter. */
+  temperature?: boolean;
+  /** Model supports tool / function calling. */
+  tool_call?: boolean;
+}
+
+/**
+ * Default per-model context window sizes (tokens) for the curated default catalog.
+ * Matches the context lengths used by OmniRoute's provider registry.
+ */
+export const OMNIROUTE_DEFAULT_MODEL_CONTEXT_LENGTHS: Record<string, number> = {
+  "cc/claude-opus-4-8": 1_000_000,
+  "cc/claude-opus-4-7": 1_000_000,
+  "cc/claude-sonnet-4-6": 200_000,
+  "cc/claude-haiku-4-5-20251001": 200_000,
+  "claude-opus-4-5-thinking": 200_000,
+  "claude-sonnet-4-5-thinking": 200_000,
+  "gemini-3.1-pro-high": 1_000_000,
+  "gemini-3-flash": 1_000_000,
+};
+
+/**
+ * Default per-model capability hints for the curated default catalog.
+ *
+ * Conservative defaults: every default model accepts attachments, tool calls
+ * and temperature; `reasoning` is opt-in per model id. Callers override per
+ * model via `OmniRouteProviderOptions.modelCapabilities`.
+ */
+export const OMNIROUTE_DEFAULT_MODEL_CAPABILITIES: Record<string, ModelCapabilities> = {
+  "cc/claude-opus-4-8": { attachment: true, reasoning: true, temperature: true, tool_call: true },
+  "cc/claude-opus-4-7": { attachment: true, reasoning: true, temperature: true, tool_call: true },
+  "cc/claude-sonnet-4-6": { attachment: true, reasoning: true, temperature: true, tool_call: true },
+  "cc/claude-haiku-4-5-20251001": { attachment: true, temperature: true, tool_call: true },
+  "claude-opus-4-5-thinking": {
+    attachment: true,
+    reasoning: true,
+    temperature: true,
+    tool_call: true,
+  },
+  "claude-sonnet-4-5-thinking": {
+    attachment: true,
+    reasoning: true,
+    temperature: true,
+    tool_call: true,
+  },
+  "gemini-3.1-pro-high": { attachment: true, reasoning: true, temperature: true, tool_call: true },
+  "gemini-3-flash": { attachment: true, temperature: true, tool_call: true },
+};
+
+export interface OmniRouteProviderOptions {
+  /** OmniRoute base URL, with or without trailing `/v1`. Required. */
+  baseURL: string;
+  /** OmniRoute API key. Required. Use `sk_omniroute` for local instances without REQUIRE_API_KEY. */
+  apiKey: string;
+  /** Override the display name shown in OpenCode. Default: `"OmniRoute"`. */
+  displayName?: string;
+  /** Override the model catalog. Accepts model ids (strings) or live model entries from `fetchLiveModels`. When entries carry a `contextLength`, it is used directly — no hardcoded map needed. */
+  models?: readonly (string | { id: string; contextLength?: number })[];
+  /** Optional human-readable labels keyed by model id. Overridden by `modelCapabilities[id].label`. */
+  modelLabels?: Record<string, string>;
+  /**
+   * Optional capability overrides keyed by model id. Merged on top of
+   * `OMNIROUTE_DEFAULT_MODEL_CAPABILITIES` for ids in the default catalog;
+   * for custom ids the override is used verbatim.
+   */
+  modelCapabilities?: Record<string, ModelCapabilities>;
+  /**
+   * Optional per-model context-length overrides (tokens). Takes precedence
+   * over the static `OMNIROUTE_DEFAULT_MODEL_CONTEXT_LENGTHS` map but is
+   * superseded by `contextLength` on live model entries passed via `models`.
+   */
+  modelContextLengths?: Record<string, string | number>;
+  /**
+   * Primary model for OpenCode (top-level `model` key).
+   * Emitted as `"omniroute/<id>"`. When omitted the key is not written.
+   */
+  model?: string;
+  /**
+   * Secondary / cheap model for OpenCode (top-level `small_model` key).
+   * Emitted as `"omniroute/<id>"`. When omitted the key is not written.
+   */
+  smallModel?: string;
+}
+
+/** Per-model entry written under `provider.omniroute.models[id]`. */
+export interface OpenCodeModelEntry {
+  name: string;
+  attachment?: boolean;
+  reasoning?: boolean;
+  temperature?: boolean;
+  tool_call?: boolean;
+  /**
+   * Context window limit. OpenCode reads this to determine usable context
+   * length for compaction, overflow detection, and router decisions.
+   * Maps to `limit.context` in OpenCode's provider config schema.
+   */
+  limit?: {
+    /** Maximum context length in tokens (e.g. 200000 for Claude, 1000000 for Gemini). */
+    context: number;
+    /** Optional per-request max input tokens. */
+    input?: number;
+    /** Optional max output tokens. */
+    output?: number;
+  };
+}
+
+export interface OpenCodeProviderEntry {
+  /** Identifier of the OpenCode runtime package that will speak to OmniRoute. */
+  npm: typeof OMNIROUTE_PROVIDER_NPM;
+  /** Display name in the OpenCode UI. */
+  name: string;
+  /** Options forwarded to `@ai-sdk/openai-compatible`. */
+  options: {
+    baseURL: string;
+    apiKey: string;
+  };
+  /** Model catalog surfaced to OpenCode. */
+  models: Record<string, OpenCodeModelEntry>;
+}
+
+export interface OpenCodeConfigDocument {
+  $schema: typeof OPENCODE_CONFIG_SCHEMA;
+  /** Primary model for OpenCode, e.g. `"omniroute/claude-sonnet-4-5-thinking"`. */
+  model?: string;
+  /** Secondary / cheap model for OpenCode, e.g. `"omniroute/gemini-3-flash"`. */
+  small_model?: string;
+  provider: {
+    [OMNIROUTE_PROVIDER_KEY]: OpenCodeProviderEntry;
+  };
+}
+
+function requireNonEmpty(value: unknown, field: string): string {
+  if (typeof value !== "string") {
+    throw new TypeError(`@omniroute/opencode-provider: ${field} must be a string`);
+  }
+  const trimmed = value.trim();
+  if (!trimmed) {
+    throw new Error(`@omniroute/opencode-provider: ${field} is required and cannot be empty`);
+  }
+  return trimmed;
+}
+
+/**
+ * Normalise the user-supplied baseURL so the final `options.baseURL` always
+ * ends in exactly one `/v1`. Accepts both `http://host` and `http://host/v1`.
+ */
+export function normalizeBaseURL(rawBaseURL: string): string {
+  const trimmed = requireNonEmpty(rawBaseURL, "baseURL");
+  try {
+    new URL(trimmed);
+  } catch {
+    throw new Error(
+      `@omniroute/opencode-provider: baseURL is not a valid URL: ${JSON.stringify(rawBaseURL)}`
+    );
+  }
+  let base = trimmed;
+  let end = base.length;
+  while (end > 0 && base[end - 1] === "/") end--;
+  base = end < base.length ? base.slice(0, end) : base;
+  if (base.endsWith("/v1")) base = base.slice(0, -3);
+  return base + "/v1";
+}
+
+/**
+ * Build the `provider.omniroute` entry for an OpenCode config document.
+ * The returned object is JSON-serialisable and safe to embed verbatim.
+ */
+export function createOmniRouteProvider(options: OmniRouteProviderOptions): OpenCodeProviderEntry {
+  const baseURL = normalizeBaseURL(options.baseURL);
+  const apiKey = requireNonEmpty(options.apiKey, "apiKey");
+
+  const modelList =
+    options.models && options.models.length > 0
+      ? [...options.models]
+      : [...OMNIROUTE_DEFAULT_OPENCODE_MODELS];
+
+  const labels = options.modelLabels ?? {};
+  const overrides = options.modelCapabilities ?? {};
+  const models: Record<string, OpenCodeModelEntry> = {};
+  const seen = new Set<string>();
+  for (const raw of modelList) {
+    const id =
+      typeof raw === "object" && raw !== null && "id" in raw && typeof (raw as any).id === "string"
+        ? (raw as { id: string }).id.trim()
+        : typeof raw === "string"
+          ? raw.trim()
+          : "";
+    if (!id || seen.has(id)) continue;
+    seen.add(id);
+    const defaults = OMNIROUTE_DEFAULT_MODEL_CAPABILITIES[id] ?? {};
+    const override = overrides[id] ?? {};
+    const merged: ModelCapabilities = { ...defaults, ...override };
+    const explicitLabel =
+      typeof merged.label === "string" && merged.label.trim()
+        ? merged.label.trim()
+        : typeof labels[id] === "string" && labels[id].trim()
+          ? labels[id].trim()
+          : id;
+    const entry: OpenCodeModelEntry = { name: explicitLabel };
+    if (typeof merged.attachment === "boolean") entry.attachment = merged.attachment;
+    if (typeof merged.reasoning === "boolean") entry.reasoning = merged.reasoning;
+    if (typeof merged.temperature === "boolean") entry.temperature = merged.temperature;
+    if (typeof merged.tool_call === "boolean") entry.tool_call = merged.tool_call;
+
+    // Context window: live model entry (from API catalog) > modelContextLengths > static defaults
+    const liveContext =
+      typeof raw === "object" && raw !== null
+        ? (raw as { contextLength?: number }).contextLength
+        : undefined;
+    const rawContextLength =
+      liveContext ??
+      options.modelContextLengths?.[id] ??
+      OMNIROUTE_DEFAULT_MODEL_CONTEXT_LENGTHS[id];
+    const contextLength =
+      typeof rawContextLength === "string" ? parseInt(rawContextLength, 10) : rawContextLength;
+    if (typeof contextLength === "number" && !isNaN(contextLength) && contextLength > 0) {
+      entry.limit = { context: contextLength };
+    }
+
+    models[id] = entry;
+  }
+
+  return {
+    npm: OMNIROUTE_PROVIDER_NPM,
+    name: options.displayName?.trim() || "OmniRoute",
+    options: { baseURL, apiKey },
+    models,
+  };
+}
+
+/**
+ * Build a full OpenCode config document (with `$schema` + `provider.omniroute`).
+ * Useful when scaffolding a fresh `opencode.json`.
+ *
+ * When `options.model` / `options.smallModel` are supplied they are emitted as
+ * top-level `model` / `small_model` keys prefixed with `"omniroute/"` so
+ * OpenCode resolves them through the configured provider.
+ */
+export function buildOmniRouteOpenCodeConfig(
+  options: OmniRouteProviderOptions
+): OpenCodeConfigDocument {
+  const doc: OpenCodeConfigDocument = {
+    $schema: OPENCODE_CONFIG_SCHEMA,
+    provider: {
+      [OMNIROUTE_PROVIDER_KEY]: createOmniRouteProvider(options),
+    },
+  };
+
+  if (options.model !== undefined) {
+    const id = options.model.trim();
+    if (id) doc.model = `${OMNIROUTE_PROVIDER_KEY}/${id}`;
+  }
+
+  if (options.smallModel !== undefined) {
+    const id = options.smallModel.trim();
+    if (id) doc.small_model = `${OMNIROUTE_PROVIDER_KEY}/${id}`;
+  }
+
+  return doc;
+}
+
+/**
+ * Merge the OmniRoute provider entry (and optional `model` / `small_model`
+ * keys) into an already-existing OpenCode config object.
+ *
+ * Performs a non-destructive merge: all top-level keys in `existing` are
+ * preserved. The `provider` map is shallow-merged so other providers already
+ * present are not removed. If `existing.provider.omniroute` already exists it
+ * is overwritten by the newly built entry.
+ *
+ * `model` and `small_model` are only written when supplied in `options`.
+ *
+ * @example
+ * ```ts
+ * const existing = JSON.parse(readFileSync("opencode.json", "utf8"));
+ * const updated = mergeIntoExistingConfig(existing, {
+ *   baseURL: "http://localhost:20128",
+ *   apiKey: "sk_omniroute",
+ *   model: "claude-sonnet-4-5-thinking",
+ * });
+ * writeFileSync("opencode.json", JSON.stringify(updated, null, 2));
+ * ```
+ */
+export function mergeIntoExistingConfig(
+  existing: Record<string, unknown>,
+  options: OmniRouteProviderOptions
+): Record<string, unknown> {
+  const partial = buildOmniRouteOpenCodeConfig(options);
+
+  const merged: Record<string, unknown> = { ...existing };
+
+  if (partial.model !== undefined) merged.model = partial.model;
+  if (partial.small_model !== undefined) merged.small_model = partial.small_model;
+
+  const existingProvider =
+    typeof existing.provider === "object" && existing.provider !== null
+      ? (existing.provider as Record<string, unknown>)
+      : {};
+
+  merged.provider = {
+    ...existingProvider,
+    [OMNIROUTE_PROVIDER_KEY]: partial.provider[OMNIROUTE_PROVIDER_KEY],
+  };
+
+  return merged;
+}
+
+/**
+ * The 7 read-only MCP scopes that allow inspection without any write access.
+ * Suitable for shared / public environments.
+ */
+export const OMNIROUTE_MCP_DEFAULT_SCOPES = [
+  "read:health",
+  "read:combos",
+  "read:quota",
+  "read:usage",
+  "read:models",
+  "read:cache",
+  "read:compression",
+] as const;
+
+export type OmniRouteMCPScope = (typeof OMNIROUTE_MCP_DEFAULT_SCOPES)[number] | string;
+
+export interface OmniRouteMCPOptions {
+  /** Absolute path to the MCP server entry point (TypeScript or compiled JS). */
+  serverPath: string;
+  /** OmniRoute API key forwarded to the MCP server as `OMNIROUTE_API_KEY`. */
+  apiKey: string;
+  /**
+   * Management API key used for management-scoped operations.
+   * When supplied it is forwarded as `OMNIROUTE_MANAGEMENT_API_KEY`.
+   */
+  managementApiKey?: string;
+  /**
+   * Comma-separated scope list passed as `OMNIROUTE_MCP_SCOPES`.
+   * When omitted `OMNIROUTE_MCP_ENFORCE_SCOPES` is not set and all scopes are
+   * available (development default). Pass an explicit list to restrict access.
+   */
+  scopes?: OmniRouteMCPScope[];
+  /**
+   * Runtime used to execute the MCP server.
+   *
+   * - `"tsx"` (default) — runs via `npx tsx` for TypeScript source files.
+   * - `"node"` — runs via `node` for compiled JS outputs.
+   */
+  runtime?: "tsx" | "node";
+}
+
+export interface OpenCodeMCPServerEntry {
+  command: string;
+  args: string[];
+  env: Record<string, string>;
+}
+
+/**
+ * Build the `mcp.servers.omniroute` entry for an OpenCode config document.
+ *
+ * @example
+ * ```ts
+ * const mcpEntry = createOmniRouteMCPEntry({
+ *   serverPath: "/home/user/.local/share/omniroute/open-sse/mcp-server/server.ts",
+ *   apiKey: "sk_omniroute",
+ *   managementApiKey: "sk_manage_...",
+ *   scopes: ["read:health", "read:combos", "execute:completions"],
+ * });
+ * // Place at config.mcp.servers.omniroute
+ * ```
+ */
+export function createOmniRouteMCPEntry(options: OmniRouteMCPOptions): OpenCodeMCPServerEntry {
+  const serverPath = requireNonEmpty(options.serverPath, "serverPath");
+  const apiKey = requireNonEmpty(options.apiKey, "apiKey");
+
+  const runtime = options.runtime ?? "tsx";
+
+  const command = runtime === "tsx" ? "npx" : "node";
+  const args = runtime === "tsx" ? ["tsx", serverPath] : [serverPath];
+
+  const env: Record<string, string> = {
+    OMNIROUTE_API_KEY: apiKey,
+  };
+
+  if (options.managementApiKey !== undefined) {
+    const mgmtKey = options.managementApiKey.trim();
+    if (mgmtKey) env.OMNIROUTE_MANAGEMENT_API_KEY = mgmtKey;
+  }
+
+  if (options.scopes !== undefined && options.scopes.length > 0) {
+    env.OMNIROUTE_MCP_ENFORCE_SCOPES = "true";
+    env.OMNIROUTE_MCP_SCOPES = options.scopes.join(",");
+  }
+
+  return { command, args, env };
+}
+
+async function fetchJSON<T>(url: string, apiKey: string, timeoutMs: number): Promise<T> {
+  const controller = new AbortController();
+  const timer = setTimeout(() => controller.abort(), timeoutMs);
+
+  try {
+    const response = await fetch(url, {
+      headers: { Authorization: `Bearer ${apiKey}` },
+      signal: controller.signal,
+    });
+
+    if (!response.ok) {
+      throw new Error(`received HTTP ${response.status}`);
+    }
+
+    return (await response.json()) as T;
+  } catch (err) {
+    const message = err instanceof Error ? err.message : String(err);
+    throw new Error(`@omniroute/opencode-provider: request to ${url} failed: ${message}`);
+  } finally {
+    clearTimeout(timer);
+  }
+}
+
+/**
+ * Lightweight model descriptor returned by `fetchLiveModels`.
+ * The shape mirrors the subset of fields that OmniRoute's `/v1/models`
+ * endpoint reliably provides across versions, normalised from both
+ * camelCase and snake_case variants used by different OmniRoute releases.
+ *
+ * Attribution: field-variant normalisation logic adapted from
+ * https://github.com/Alph4d0g/opencode-omniroute-auth (MIT).
+ */
+export interface OmniRouteLiveModel {
+  id: string;
+  name: string;
+  /** Context window length in tokens (e.g. 200000 for Claude, 1000000 for Gemini). */
+  contextLength?: number;
+}
+
+/**
+ * Fetch the live model catalog from a running OmniRoute instance.
+ *
+ * Returns an array of `{ id, name }` objects from `GET /v1/models`. Handles
+ * both the camelCase (`modelId`, `displayName`) and snake_case (`model_id`,
+ * `display_name`) field variants across OmniRoute versions.
+ *
+ * Useful for dynamically populating the `models` option of
+ * `createOmniRouteProvider` / `buildOmniRouteOpenCodeConfig` instead of
+ * relying on `OMNIROUTE_DEFAULT_OPENCODE_MODELS`.
+ *
+ * @param baseURL   - OmniRoute base URL (with or without `/v1`).
+ * @param apiKey    - OmniRoute API key.
+ * @param timeoutMs - Request timeout in milliseconds (default 5000).
+ *
+ * @example
+ * ```ts
+ * const models = await fetchLiveModels("http://localhost:20128", "sk_omniroute");
+ * const config = buildOmniRouteOpenCodeConfig({
+ *   baseURL: "http://localhost:20128",
+ *   apiKey: "sk_omniroute",
+ *   models,                    // OmniRouteLiveModel[] — contextLength auto-extracted
+ *   modelLabels: Object.fromEntries(models.map((m) => [m.id, m.name])),
+ * });
+ * ```
+ */
+export async function fetchLiveModels(
+  baseURL: string,
+  apiKey: string,
+  timeoutMs = 5_000
+): Promise<OmniRouteLiveModel[]> {
+  const key = requireNonEmpty(apiKey, "apiKey");
+  const url = `${normalizeBaseURL(baseURL)}/models`;
+
+  const body = await fetchJSON<unknown>(url, key, timeoutMs);
+
+  const rawList: unknown[] = Array.isArray(body)
+    ? body
+    : body && typeof body === "object" && Array.isArray((body as { data?: unknown[] }).data)
+      ? ((body as { data: unknown[] }).data as unknown[])
+      : [];
+
+  const models: OmniRouteLiveModel[] = [];
+  for (const raw of rawList) {
+    if (typeof raw !== "object" || raw === null) continue;
+    const r = raw as Record<string, unknown>;
+
+    const id =
+      typeof r.id === "string"
+        ? r.id.trim()
+        : typeof r.modelId === "string"
+          ? r.modelId.trim()
+          : typeof r.model_id === "string"
+            ? r.model_id.trim()
+            : "";
+
+    if (!id) continue;
+
+    const name =
+      typeof r.name === "string"
+        ? r.name.trim()
+        : typeof r.displayName === "string"
+          ? r.displayName.trim()
+          : typeof r.display_name === "string"
+            ? r.display_name.trim()
+            : id;
+
+    // Extract context_length from OmniRoute's /v1/models response.
+    // OmniRoute returns context_length in snake_case for both synced
+    // models (with inputTokenLimit) and custom models; the catalog's
+    // getDefaultContextFallback also injects it from registry defaults.
+    const contextLength =
+      typeof r.context_length === "number" && r.context_length > 0
+        ? r.context_length
+        : typeof r.max_context_window_tokens === "number" && r.max_context_window_tokens > 0
+          ? r.max_context_window_tokens
+          : undefined;
+
+    models.push({ id, name: name || id, ...(contextLength ? { contextLength } : {}) });
+  }
+
+  return models;
+}
+
+/**
+ * Valid per-combo compression override values.
+ * An empty string clears any existing override (inherits global setting).
+ */
+export type OmniRouteCompressionOverride =
+  | ""
+  | "off"
+  | "lite"
+  | "standard"
+  | "aggressive"
+  | "ultra"
+  | "rtk"
+  | "stacked";
+
+const VALID_COMPRESSION_OVERRIDES = new Set<string>([
+  "",
+  "off",
+  "lite",
+  "standard",
+  "aggressive",
+  "ultra",
+  "rtk",
+  "stacked",
+]);
+
+/** Slim combo descriptor returned by `listCombos`. */
+export interface OmniRouteCombo {
+  id: string;
+  name: string;
+  strategy: string;
+  active: boolean;
+  compressionOverride: OmniRouteCompressionOverride;
+}
+
+/**
+ * Fetch the active routing combo list from a running OmniRoute instance.
+ *
+ * Returns an array of combo descriptors from `GET /api/combos`. The
+ * `compressionOverride` field reflects the per-combo compression strategy
+ * (one of the 8 recognised values; empty string means "inherit global").
+ *
+ * Requires a management-scoped API key (Bearer `manage` scope) when the
+ * instance has `REQUIRE_API_KEY` enabled.
+ *
+ * @param baseURL          - OmniRoute base URL (with or without `/v1`).
+ * @param managementApiKey - API key with `manage` scope.
+ * @param timeoutMs        - Request timeout in milliseconds (default 5000).
+ */
+export async function listCombos(
+  baseURL: string,
+  managementApiKey: string,
+  timeoutMs = 5_000
+): Promise<OmniRouteCombo[]> {
+  const key = requireNonEmpty(managementApiKey, "managementApiKey");
+  const base = normalizeBaseURL(baseURL).replace(/\/v1$/, "");
+  const url = `${base}/api/combos`;
+
+  const body = await fetchJSON<unknown>(url, key, timeoutMs);
+  const rawList: unknown[] = Array.isArray(body)
+    ? body
+    : body && typeof body === "object" && Array.isArray((body as { combos?: unknown[] }).combos)
+      ? ((body as { combos: unknown[] }).combos as unknown[])
+      : [];
+
+  const combos: OmniRouteCombo[] = [];
+  for (const raw of rawList) {
+    if (typeof raw !== "object" || raw === null) continue;
+    const r = raw as Record<string, unknown>;
+
+    const id = typeof r.id === "string" ? r.id.trim() : "";
+    if (!id) continue;
+
+    const name = typeof r.name === "string" ? r.name.trim() : id;
+    const strategy = typeof r.strategy === "string" ? r.strategy : "";
+    const active = typeof r.active === "boolean" ? r.active : false;
+
+    const rawOverride = typeof r.compressionOverride === "string" ? r.compressionOverride : "";
+    const compressionOverride = VALID_COMPRESSION_OVERRIDES.has(rawOverride)
+      ? (rawOverride as OmniRouteCompressionOverride)
+      : "";
+
+    combos.push({ id, name, strategy, active, compressionOverride });
+  }
+
+  return combos;
+}
+
+/**
+ * Options for `createOmniRouteComboConfig`.
+ * Mirrors the subset of combo fields exposed by the OmniRoute `/api/combos`
+ * PATCH / POST payload that are safe to set programmatically.
+ */
+export interface OmniRouteComboConfigOptions {
+  /** Human-readable combo name. */
+  name: string;
+  /** Routing strategy (e.g. `"priority"`, `"weighted"`, `"round-robin"`). */
+  strategy: string;
+  /**
+   * Per-combo compression override.
+   * Empty string removes any override (inherits global setting).
+   */
+  compressionOverride?: OmniRouteCompressionOverride;
+  /** Whether this combo is active for routing. Default: `true`. */
+  active?: boolean;
+  /**
+   * Ordered list of provider IDs in this combo.
+   * Required for create operations; optional for updates.
+   */
+  providers?: string[];
+}
+
+/**
+ * Build a typed combo payload suitable for OmniRoute's management API.
+ *
+ * The returned object is JSON-serialisable and safe to pass as the body of a
+ * `POST /api/combos` (create) or `PATCH /api/combos/:id` (update) request.
+ *
+ * @example
+ * ```ts
+ * const payload = createOmniRouteComboConfig({
+ *   name: "claude-primary",
+ *   strategy: "priority",
+ *   compressionOverride: "standard",
+ *   providers: ["anthropic-claude-opus", "anthropic-claude-sonnet"],
+ * });
+ * await fetch(`${baseURL}/api/combos`, {
+ *   method: "POST",
+ *   headers: { Authorization: `Bearer ${mgmtKey}`, "Content-Type": "application/json" },
+ *   body: JSON.stringify(payload),
+ * });
+ * ```
+ */
+export function createOmniRouteComboConfig(
+  options: OmniRouteComboConfigOptions
+): Record<string, unknown> {
+  const name = requireNonEmpty(options.name, "name");
+  const strategy = requireNonEmpty(options.strategy, "strategy");
+
+  const payload: Record<string, unknown> = {
+    name,
+    strategy,
+    active: options.active ?? true,
+  };
+
+  if (options.compressionOverride !== undefined) {
+    payload.compressionOverride = options.compressionOverride;
+  }
+
+  if (options.providers !== undefined) {
+    const providers = options.providers.filter((p) => typeof p === "string" && p.trim());
+    if (providers.length > 0) {
+      payload.providers = providers;
+    }
+  }
+
+  return payload;
+}
+
+/**
+ * Override fields supported per agent / mode entry. Mirrors the subset of
+ * OpenCode's `AgentConfig` schema that is safe to set declaratively from a
+ * config generator. Only fields present in
+ * https://opencode.ai/config.json#AgentConfig are exposed.
+ */
+export interface OmniRouteRoleOverrides {
+  /** Forward to OpenCode's `temperature` field. */
+  temperature?: number;
+  /** Forward to OpenCode's `top_p` field. */
+  top_p?: number;
+}
+
+/** Per-role binding used by `createOmniRouteAgentBlock`. */
+export interface OmniRouteAgentRole extends OmniRouteRoleOverrides {
+  /** OmniRoute model id, e.g. `"claude-sonnet-4-5-thinking"`. */
+  modelId: string;
+  /** Optional tools allow-list; per OpenCode schema, map of tool name → enabled. */
+  tools?: Record<string, boolean>;
+  /** Optional system prompt for this agent role. */
+  prompt?: string;
+}
+
+/** Options for `createOmniRouteAgentBlock`. */
+export interface OmniRouteAgentBlockOptions {
+  /** Per-role bindings. Keys become entries under OpenCode's `agent` block. */
+  roles: Record<string, OmniRouteAgentRole>;
+}
+
+/** Single entry inside the emitted OpenCode `agent` block. */
+export interface OpenCodeAgentEntry extends OmniRouteRoleOverrides {
+  /** Always emitted as `"omniroute/<modelId>"`. */
+  model: string;
+  /** Per OpenCode schema, `Record<string, boolean>`. */
+  tools?: Record<string, boolean>;
+  /** Optional system prompt. */
+  prompt?: string;
+}
+
+function buildAgentEntry(role: OmniRouteAgentRole): OpenCodeAgentEntry | undefined {
+  if (!role || typeof role.modelId !== "string") return undefined;
+  const modelId = role.modelId.trim();
+  if (!modelId) return undefined;
+  const entry: OpenCodeAgentEntry = { model: `${OMNIROUTE_PROVIDER_KEY}/${modelId}` };
+  if (typeof role.temperature === "number") entry.temperature = role.temperature;
+  if (typeof role.top_p === "number") entry.top_p = role.top_p;
+  if (role.tools && typeof role.tools === "object" && !Array.isArray(role.tools)) {
+    const tools: Record<string, boolean> = {};
+    for (const [name, enabled] of Object.entries(role.tools)) {
+      if (typeof name !== "string" || !name.trim()) continue;
+      if (typeof enabled !== "boolean") continue;
+      tools[name] = enabled;
+    }
+    if (Object.keys(tools).length > 0) entry.tools = tools;
+  }
+  if (typeof role.prompt === "string" && role.prompt.trim()) {
+    entry.prompt = role.prompt;
+  }
+  return entry;
+}
+
+/**
+ * Build the OpenCode `agent` block, pre-wired so each agent role routes to a
+ * specific OmniRoute model. Useful for `.opencode/agent/*.md` defaults and
+ * scaffolded `opencode.json` files.
+ *
+ * Emitted fields are limited to those declared in OpenCode's `AgentConfig`
+ * schema (`model`, `temperature`, `top_p`, `tools`, `prompt`). The `tools`
+ * field is a `Record<string, boolean>` per the schema, not a string array.
+ *
+ * Roles with empty / missing `modelId` are skipped.
+ *
+ * @example
+ * ```ts
+ * const agentBlock = createOmniRouteAgentBlock({
+ *   roles: {
+ *     build: { modelId: "claude-sonnet-4-5-thinking", temperature: 0.2 },
+ *     plan: { modelId: "claude-opus-4-5-thinking", top_p: 0.95 },
+ *     review: { modelId: "gemini-3-flash", tools: { edit: false, bash: false } },
+ *   },
+ * });
+ * // -> { build: { model: "omniroute/claude-sonnet-4-5-thinking", temperature: 0.2 }, ... }
+ * ```
+ */
+export function createOmniRouteAgentBlock(
+  options: OmniRouteAgentBlockOptions
+): Record<string, OpenCodeAgentEntry> {
+  const out: Record<string, OpenCodeAgentEntry> = {};
+  const roles = options.roles ?? {};
+  for (const [roleName, role] of Object.entries(roles)) {
+    const entry = buildAgentEntry(role);
+    if (entry) out[roleName] = entry;
+  }
+  return out;
+}
+
+/**
+ * Per-mode binding used by `createOmniRouteModesBlock`.
+ *
+ * @deprecated OpenCode's top-level `mode` block is deprecated in favour of
+ * `agent`. Prefer `OmniRouteAgentRole` + `createOmniRouteAgentBlock`. This
+ * type and the corresponding helper are kept for back-compat with configs
+ * still using `mode:`.
+ */
+export interface OmniRouteMode extends OmniRouteAgentRole {}
+
+/**
+ * Options for `createOmniRouteModesBlock`.
+ *
+ * @deprecated See `OmniRouteMode`.
+ */
+export interface OmniRouteModesBlockOptions {
+  /** Per-mode bindings. Keys become entries under OpenCode's deprecated top-level `mode` block. */
+  modes: Record<string, OmniRouteMode>;
+}
+
+/**
+ * Single entry inside the emitted OpenCode `mode` block.
+ *
+ * @deprecated See `OmniRouteMode`.
+ */
+export interface OpenCodeModeEntry extends OpenCodeAgentEntry {}
+
+/**
+ * Build the OpenCode top-level `mode` block, pre-wired so each mode routes to
+ * a specific OmniRoute model. Emits the same shape as the `agent` block since
+ * OpenCode's schema treats them identically (both reference `AgentConfig`).
+ *
+ * Modes with empty / missing `modelId` are skipped.
+ *
+ * @deprecated OpenCode's top-level `mode` block is deprecated in favour of
+ * `agent`. Prefer `createOmniRouteAgentBlock`. This helper is kept for
+ * back-compat with configs still using `mode:`.
+ *
+ * @example
+ * ```ts
+ * const modesBlock = createOmniRouteModesBlock({
+ *   modes: {
+ *     build: { modelId: "claude-sonnet-4-5-thinking", tools: { edit: true, bash: true } },
+ *     plan: { modelId: "claude-opus-4-5-thinking", prompt: "Plan first, code later." },
+ *     review: { modelId: "gemini-3-flash" },
+ *   },
+ * });
+ * ```
+ */
+export function createOmniRouteModesBlock(
+  options: OmniRouteModesBlockOptions
+): Record<string, OpenCodeModeEntry> {
+  const out: Record<string, OpenCodeModeEntry> = {};
+  const modes = options.modes ?? {};
+  for (const [modeName, mode] of Object.entries(modes)) {
+    const entry = buildAgentEntry(mode);
+    if (entry) out[modeName] = entry;
+  }
+  return out;
+}
+
+export default createOmniRouteProvider;

@omniroute/opencode-provider/tests/index.test.ts

@@ -0,0 +1,686 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+import { createServer } from "node:http";
+import type { Server } from "node:http";
+import { readFileSync } from "node:fs";
+import { fileURLToPath } from "node:url";
+import { dirname, join } from "node:path";
+
+import {
+  buildOmniRouteOpenCodeConfig,
+  createOmniRouteAgentBlock,
+  createOmniRouteComboConfig,
+  createOmniRouteMCPEntry,
+  createOmniRouteModesBlock,
+  createOmniRouteProvider,
+  fetchLiveModels,
+  listCombos,
+  mergeIntoExistingConfig,
+  normalizeBaseURL,
+  OMNIROUTE_DEFAULT_MODEL_CAPABILITIES,
+  OMNIROUTE_DEFAULT_MODEL_CONTEXT_LENGTHS,
+  OMNIROUTE_DEFAULT_OPENCODE_MODELS,
+  OMNIROUTE_MCP_DEFAULT_SCOPES,
+  OMNIROUTE_PROVIDER_NPM,
+  OPENCODE_CONFIG_SCHEMA,
+} from "../src/index.ts";
+
+test("normalizeBaseURL preserves a bare host:port", () => {
+  assert.equal(normalizeBaseURL("http://localhost:20128"), "http://localhost:20128/v1");
+});
+
+test("normalizeBaseURL strips trailing slashes", () => {
+  assert.equal(normalizeBaseURL("http://localhost:20128////"), "http://localhost:20128/v1");
+});
+
+test("normalizeBaseURL deduplicates an existing /v1 suffix", () => {
+  assert.equal(normalizeBaseURL("http://localhost:20128/v1"), "http://localhost:20128/v1");
+  assert.equal(normalizeBaseURL("http://localhost:20128/v1/"), "http://localhost:20128/v1");
+});
+
+test("normalizeBaseURL rejects empty input", () => {
+  assert.throws(() => normalizeBaseURL("   "), /baseURL is required/);
+});
+
+test("normalizeBaseURL rejects malformed URLs", () => {
+  assert.throws(() => normalizeBaseURL("not a url"), /not a valid URL/);
+});
+
+test("createOmniRouteProvider validates required fields", () => {
+  assert.throws(
+    () => createOmniRouteProvider({ baseURL: "", apiKey: "x" } as never),
+    /baseURL is required/
+  );
+  assert.throws(
+    () => createOmniRouteProvider({ baseURL: "http://x", apiKey: "" } as never),
+    /apiKey is required/
+  );
+});
+
+test("createOmniRouteProvider produces the OpenCode-compatible shape", () => {
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+  });
+
+  assert.equal(provider.npm, OMNIROUTE_PROVIDER_NPM);
+  assert.equal(provider.name, "OmniRoute");
+  assert.equal(provider.options.baseURL, "http://localhost:20128/v1");
+  assert.equal(provider.options.apiKey, "sk_omniroute");
+  assert.equal(typeof provider.models, "object");
+});
+
+test("createOmniRouteProvider seeds the default model catalog", () => {
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+  });
+
+  const modelIds = Object.keys(provider.models).sort();
+  const defaultIds = [...OMNIROUTE_DEFAULT_OPENCODE_MODELS].sort();
+  assert.deepEqual(modelIds, defaultIds);
+  for (const id of defaultIds) {
+    assert.equal(provider.models[id]?.name, id);
+    assert.equal(provider.models[id]?.attachment, true);
+  }
+});
+
+test("createOmniRouteProvider honours a custom models list and labels", () => {
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+    models: ["auto", "claude-opus-4-7"],
+    modelLabels: { auto: "Auto-Combo", "claude-opus-4-7": "Opus 4.7" },
+  });
+
+  assert.deepEqual(Object.keys(provider.models), ["auto", "claude-opus-4-7"]);
+  assert.equal(provider.models.auto.name, "Auto-Combo");
+  assert.equal(provider.models["claude-opus-4-7"].name, "Opus 4.7");
+});
+
+test("createOmniRouteProvider deduplicates and trims model ids", () => {
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+    models: ["  auto  ", "auto", "", "claude-opus-4-7"],
+  });
+  assert.deepEqual(Object.keys(provider.models), ["auto", "claude-opus-4-7"]);
+});
+
+test("createOmniRouteProvider honours displayName override", () => {
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+    displayName: "Local OmniRoute",
+  });
+  assert.equal(provider.name, "Local OmniRoute");
+});
+
+test("buildOmniRouteOpenCodeConfig wraps the provider with the OpenCode schema", () => {
+  const doc = buildOmniRouteOpenCodeConfig({
+    baseURL: "http://localhost:20128/v1",
+    apiKey: "sk_omniroute",
+  });
+
+  assert.equal(doc.$schema, OPENCODE_CONFIG_SCHEMA);
+  assert.equal(typeof doc.provider.omniroute, "object");
+  assert.equal(doc.provider.omniroute.options.baseURL, "http://localhost:20128/v1");
+});
+
+test("config document is JSON-serialisable", () => {
+  const doc = buildOmniRouteOpenCodeConfig({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+  });
+  const round = JSON.parse(JSON.stringify(doc));
+  assert.deepEqual(round, doc);
+});
+
+test("buildOmniRouteOpenCodeConfig emits model and small_model prefixed with provider key", () => {
+  const doc = buildOmniRouteOpenCodeConfig({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+    model: "claude-sonnet-4-5-thinking",
+    smallModel: "gemini-3-flash",
+  });
+  assert.equal(doc.model, "omniroute/claude-sonnet-4-5-thinking");
+  assert.equal(doc.small_model, "omniroute/gemini-3-flash");
+});
+
+test("buildOmniRouteOpenCodeConfig omits model and small_model when not supplied", () => {
+  const doc = buildOmniRouteOpenCodeConfig({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+  });
+  assert.equal(doc.model, undefined);
+  assert.equal(doc.small_model, undefined);
+  assert.ok(!("model" in doc));
+  assert.ok(!("small_model" in doc));
+});
+
+test("buildOmniRouteOpenCodeConfig ignores blank model strings", () => {
+  const doc = buildOmniRouteOpenCodeConfig({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+    model: "   ",
+    smallModel: "",
+  });
+  assert.ok(!("model" in doc));
+  assert.ok(!("small_model" in doc));
+});
+
+test("mergeIntoExistingConfig preserves existing provider entries", () => {
+  const existing = {
+    $schema: OPENCODE_CONFIG_SCHEMA,
+    provider: {
+      anthropic: { npm: "@ai-sdk/anthropic", name: "Anthropic", options: {}, models: {} },
+    },
+    keybinds: { submit: "enter" },
+  };
+  const result = mergeIntoExistingConfig(existing, {
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+  });
+  assert.ok("anthropic" in (result.provider as Record<string, unknown>));
+  assert.ok("omniroute" in (result.provider as Record<string, unknown>));
+  assert.deepEqual((result as Record<string, unknown>).keybinds, { submit: "enter" });
+});
+
+test("mergeIntoExistingConfig overwrites existing omniroute entry", () => {
+  const existing = {
+    provider: {
+      omniroute: {
+        npm: "@ai-sdk/openai-compatible",
+        name: "OLD",
+        options: { baseURL: "http://old/v1", apiKey: "old" },
+        models: {},
+      },
+    },
+  };
+  const result = mergeIntoExistingConfig(existing, {
+    baseURL: "http://new",
+    apiKey: "new-key",
+    displayName: "NEW",
+  });
+  const omniroute = (result.provider as Record<string, unknown>).omniroute as { name: string };
+  assert.equal(omniroute.name, "NEW");
+});
+
+test("mergeIntoExistingConfig writes model and small_model when supplied", () => {
+  const result = mergeIntoExistingConfig(
+    {},
+    {
+      baseURL: "http://localhost:20128",
+      apiKey: "sk_omniroute",
+      model: "claude-sonnet-4-5-thinking",
+      smallModel: "gemini-3-flash",
+    }
+  );
+  assert.equal(result.model, "omniroute/claude-sonnet-4-5-thinking");
+  assert.equal(result.small_model, "omniroute/gemini-3-flash");
+});
+
+test("mergeIntoExistingConfig does not add model keys when not supplied", () => {
+  const result = mergeIntoExistingConfig(
+    {},
+    { baseURL: "http://localhost:20128", apiKey: "sk_omniroute" }
+  );
+  assert.ok(!("model" in result));
+  assert.ok(!("small_model" in result));
+});
+
+test("OMNIROUTE_MCP_DEFAULT_SCOPES contains 7 read-only scopes", () => {
+  assert.equal(OMNIROUTE_MCP_DEFAULT_SCOPES.length, 7);
+  assert.ok(OMNIROUTE_MCP_DEFAULT_SCOPES.every((s) => s.startsWith("read:")));
+});
+
+test("createOmniRouteMCPEntry defaults to tsx runtime", () => {
+  const entry = createOmniRouteMCPEntry({
+    serverPath: "/path/to/server.ts",
+    apiKey: "sk_omniroute",
+  });
+  assert.equal(entry.command, "npx");
+  assert.deepEqual(entry.args, ["tsx", "/path/to/server.ts"]);
+  assert.equal(entry.env.OMNIROUTE_API_KEY, "sk_omniroute");
+  assert.ok(!("OMNIROUTE_MCP_ENFORCE_SCOPES" in entry.env));
+  assert.ok(!("OMNIROUTE_MANAGEMENT_API_KEY" in entry.env));
+});
+
+test("createOmniRouteMCPEntry uses node runtime when specified", () => {
+  const entry = createOmniRouteMCPEntry({
+    serverPath: "/path/to/server.js",
+    apiKey: "sk_omniroute",
+    runtime: "node",
+  });
+  assert.equal(entry.command, "node");
+  assert.deepEqual(entry.args, ["/path/to/server.js"]);
+});
+
+test("createOmniRouteMCPEntry sets management key and scopes when supplied", () => {
+  const entry = createOmniRouteMCPEntry({
+    serverPath: "/path/to/server.ts",
+    apiKey: "sk_omniroute",
+    managementApiKey: "sk_manage",
+    scopes: ["read:health", "read:combos", "execute:completions"],
+  });
+  assert.equal(entry.env.OMNIROUTE_MANAGEMENT_API_KEY, "sk_manage");
+  assert.equal(entry.env.OMNIROUTE_MCP_ENFORCE_SCOPES, "true");
+  assert.equal(entry.env.OMNIROUTE_MCP_SCOPES, "read:health,read:combos,execute:completions");
+});
+
+test("createOmniRouteMCPEntry rejects missing required fields", () => {
+  assert.throws(
+    () => createOmniRouteMCPEntry({ serverPath: "", apiKey: "x" }),
+    /serverPath is required/
+  );
+  assert.throws(
+    () => createOmniRouteMCPEntry({ serverPath: "/p", apiKey: "" }),
+    /apiKey is required/
+  );
+});
+
+function startMockServer(
+  handler: (path: string) => unknown
+): Promise<{ url: string; close: () => void }> {
+  return new Promise((resolve) => {
+    const server: Server = createServer((req, res) => {
+      const body = JSON.stringify(handler(req.url ?? ""));
+      res.writeHead(200, { "Content-Type": "application/json" });
+      res.end(body);
+    });
+    server.listen(0, "127.0.0.1", () => {
+      const addr = server.address() as { port: number };
+      resolve({ url: `http://127.0.0.1:${addr.port}`, close: () => server.close() });
+    });
+  });
+}
+
+test("fetchLiveModels handles array envelope", async () => {
+  const { url, close } = await startMockServer(() => [
+    { id: "claude-sonnet", name: "Claude Sonnet" },
+    { id: "gemini-flash", displayName: "Gemini Flash" },
+  ]);
+  try {
+    const models = await fetchLiveModels(url, "sk_test");
+    assert.equal(models.length, 2);
+    assert.equal(models[0].id, "claude-sonnet");
+    assert.equal(models[0].name, "Claude Sonnet");
+    assert.equal(models[1].id, "gemini-flash");
+    assert.equal(models[1].name, "Gemini Flash");
+  } finally {
+    close();
+  }
+});
+
+test("fetchLiveModels handles data-envelope and snake_case fields", async () => {
+  const { url, close } = await startMockServer(() => ({
+    data: [{ model_id: "gpt-4o", display_name: "GPT-4o" }],
+  }));
+  try {
+    const models = await fetchLiveModels(url, "sk_test");
+    assert.equal(models.length, 1);
+    assert.equal(models[0].id, "gpt-4o");
+    assert.equal(models[0].name, "GPT-4o");
+  } finally {
+    close();
+  }
+});
+
+test("fetchLiveModels falls back to id as name when no name field", async () => {
+  const { url, close } = await startMockServer(() => [{ id: "auto" }]);
+  try {
+    const models = await fetchLiveModels(url, "sk_test");
+    assert.equal(models[0].name, "auto");
+  } finally {
+    close();
+  }
+});
+
+test("listCombos normalises compressionOverride", async () => {
+  const { url, close } = await startMockServer(() => ({
+    combos: [
+      {
+        id: "c1",
+        name: "Primary",
+        strategy: "priority",
+        active: true,
+        compressionOverride: "standard",
+      },
+      {
+        id: "c2",
+        name: "Cheap",
+        strategy: "weighted",
+        active: false,
+        compressionOverride: "unknown-value",
+      },
+      { id: "c3", name: "Off", strategy: "round-robin", active: true, compressionOverride: "" },
+    ],
+  }));
+  try {
+    const combos = await listCombos(url, "sk_manage");
+    assert.equal(combos.length, 3);
+    assert.equal(combos[0].compressionOverride, "standard");
+    assert.equal(combos[1].compressionOverride, "");
+    assert.equal(combos[2].compressionOverride, "");
+  } finally {
+    close();
+  }
+});
+
+test("createOmniRouteComboConfig builds minimal payload", () => {
+  const payload = createOmniRouteComboConfig({ name: "my-combo", strategy: "priority" });
+  assert.equal(payload.name, "my-combo");
+  assert.equal(payload.strategy, "priority");
+  assert.equal(payload.active, true);
+  assert.ok(!("compressionOverride" in payload));
+  assert.ok(!("providers" in payload));
+});
+
+test("createOmniRouteComboConfig includes optional fields when supplied", () => {
+  const payload = createOmniRouteComboConfig({
+    name: "full",
+    strategy: "weighted",
+    compressionOverride: "aggressive",
+    active: false,
+    providers: ["provider-a", "provider-b"],
+  });
+  assert.equal(payload.compressionOverride, "aggressive");
+  assert.equal(payload.active, false);
+  assert.deepEqual(payload.providers, ["provider-a", "provider-b"]);
+});
+
+test("OMNIROUTE_DEFAULT_OPENCODE_MODELS includes cc/ prefixed models", () => {
+  const defaults = [...OMNIROUTE_DEFAULT_OPENCODE_MODELS];
+  assert.ok(defaults.includes("cc/claude-opus-4-8"));
+  assert.ok(
+    defaults.some((m) => m.startsWith("cc/")),
+    "should have cc/ prefixed models"
+  );
+  assert.ok(defaults.length >= 7, "should have at least 7 models");
+});
+
+test("OMNIROUTE_DEFAULT_MODEL_CONTEXT_LENGTHS covers every default model id", () => {
+  for (const id of OMNIROUTE_DEFAULT_OPENCODE_MODELS) {
+    const ctx = OMNIROUTE_DEFAULT_MODEL_CONTEXT_LENGTHS[id];
+    assert.ok(
+      typeof ctx === "number" && ctx > 0,
+      `default context_length for ${id} missing — should be a positive number`
+    );
+    // Sanity: context should be at least 8K, at most 2M tokens
+    assert.ok(ctx >= 8_000, `${id} context_length ${ctx} seems too low`);
+    assert.ok(ctx <= 2_000_000, `${id} context_length ${ctx} seems too high`);
+  }
+});
+
+test("createOmniRouteProvider emits limit.context on default model entries", () => {
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+  });
+  const entry = provider.models["cc/claude-opus-4-8"];
+  assert.ok(entry.limit, "model entry should have a limit field");
+  assert.equal(entry.limit!.context, 1_000_000);
+  assert.equal(provider.models["cc/claude-opus-4-7"].limit!.context, 1_000_000);
+});
+
+test("createOmniRouteProvider omits limit.context for unknown model ids", () => {
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+    models: ["completely-unknown-model"],
+  });
+  const entry = provider.models["completely-unknown-model"];
+  assert.equal(entry.limit, undefined);
+});
+
+test("createOmniRouteProvider reads contextLength from a live model entry for ids absent from the static map", () => {
+  // #3298 regression guard: the static OMNIROUTE_DEFAULT_MODEL_CONTEXT_LENGTHS
+  // map only covers the legacy 8 Claude/Gemini ids. Before this change, any
+  // other model got `undefined` context (see the test above, string form) and
+  // OpenCode silently fell back to its 128K internal default. A live model
+  // entry carrying `contextLength` must now surface as `limit.context`.
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+    models: [{ id: "completely-unknown-model", contextLength: 262_144 }],
+  });
+  const entry = provider.models["completely-unknown-model"];
+  assert.ok(entry.limit, "a live contextLength should produce a limit field even for ids absent from the static map");
+  assert.equal(entry.limit!.context, 262_144);
+});
+
+test("createOmniRouteProvider: a live model contextLength wins over the static default map", () => {
+  // `cc/claude-opus-4-8` has a static default (1_000_000). A live entry carrying
+  // a different contextLength must take precedence (live > modelContextLengths >
+  // static defaults).
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+    models: [{ id: "cc/claude-opus-4-8", contextLength: 524_288 }],
+  });
+  assert.equal(provider.models["cc/claude-opus-4-8"].limit!.context, 524_288);
+});
+
+test("createOmniRouteProvider serialises limit.context to JSON", () => {
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+  });
+  const round = JSON.parse(JSON.stringify(provider));
+  for (const id of OMNIROUTE_DEFAULT_OPENCODE_MODELS) {
+    const expectedContext = OMNIROUTE_DEFAULT_MODEL_CONTEXT_LENGTHS[id];
+    assert.equal(
+      round.models[id].limit?.context,
+      expectedContext,
+      `${id} should serialise limit.context=${expectedContext}`
+    );
+  }
+});
+
+test("fetchLiveModels extracts context_length from snake_case field", async () => {
+  const { url, close } = await startMockServer(() => ({
+    data: [
+      { id: "cc/claude-opus-4-7", name: "Claude Opus 4.7", context_length: 200_000 },
+      { id: "gemini-3.1-pro-high", name: "Gemini 3.1 Pro", context_length: 1_000_000 },
+      { id: "no-context", name: "No Context" },
+    ],
+  }));
+  try {
+    const models = await fetchLiveModels(url, "sk_test");
+    const claude = models.find((m) => m.id === "cc/claude-opus-4-7");
+    assert.ok(claude, "claude model should be present");
+    assert.equal(claude!.contextLength, 200_000);
+    const gemini = models.find((m) => m.id === "gemini-3.1-pro-high");
+    assert.equal(gemini!.contextLength, 1_000_000);
+    const noCtx = models.find((m) => m.id === "no-context");
+    assert.equal(noCtx!.contextLength, undefined);
+  } finally {
+    close();
+  }
+});
+
+test("OMNIROUTE_DEFAULT_MODEL_CAPABILITIES covers every default model id", () => {
+  for (const id of OMNIROUTE_DEFAULT_OPENCODE_MODELS) {
+    const caps = OMNIROUTE_DEFAULT_MODEL_CAPABILITIES[id];
+    assert.ok(caps, `default capabilities for ${id} missing`);
+    assert.equal(caps.attachment, true, `${id} should default to attachment=true`);
+    assert.equal(caps.tool_call, true, `${id} should default to tool_call=true`);
+  }
+});
+
+test("createOmniRouteProvider emits default capability flags inline with the model entry", () => {
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+  });
+  const entry = provider.models["cc/claude-opus-4-8"];
+  assert.equal(entry.name, "cc/claude-opus-4-8");
+  assert.equal(entry.attachment, true);
+  assert.equal(entry.reasoning, true);
+  assert.equal(entry.temperature, true);
+  assert.equal(entry.tool_call, true);
+});
+
+test("createOmniRouteProvider modelCapabilities overrides defaults and merges per id", () => {
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+    modelCapabilities: {
+      "cc/claude-opus-4-7": { reasoning: false, label: "Opus (no thinking)" },
+    },
+  });
+  const entry = provider.models["cc/claude-opus-4-7"];
+  assert.equal(entry.name, "Opus (no thinking)");
+  assert.equal(entry.reasoning, false);
+  assert.equal(entry.attachment, true);
+  assert.equal(entry.tool_call, true);
+});
+
+test("createOmniRouteProvider applies capability overrides to non-default model ids", () => {
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+    models: ["custom-model"],
+    modelCapabilities: {
+      "custom-model": { attachment: false, tool_call: true, label: "Custom" },
+    },
+  });
+  const entry = provider.models["custom-model"];
+  assert.equal(entry.name, "Custom");
+  assert.equal(entry.attachment, false);
+  assert.equal(entry.tool_call, true);
+  assert.equal(entry.reasoning, undefined);
+  assert.equal(entry.temperature, undefined);
+});
+
+test("createOmniRouteProvider modelLabels still works when modelCapabilities omits label", () => {
+  const provider = createOmniRouteProvider({
+    baseURL: "http://localhost:20128",
+    apiKey: "sk_omniroute",
+    models: ["claude-opus-4-5-thinking"],
+    modelLabels: { "claude-opus-4-5-thinking": "Opus 4.5 (legacy label)" },
+  });
+  assert.equal(provider.models["claude-opus-4-5-thinking"].name, "Opus 4.5 (legacy label)");
+});
+
+test("createOmniRouteAgentBlock builds provider-prefixed entries per role", () => {
+  const block = createOmniRouteAgentBlock({
+    roles: {
+      build: { modelId: "claude-sonnet-4-5-thinking", temperature: 0.2 },
+      plan: { modelId: "claude-opus-4-5-thinking", top_p: 0.95 },
+      review: { modelId: "gemini-3-flash", temperature: 0.0 },
+    },
+  });
+  assert.equal(block.build.model, "omniroute/claude-sonnet-4-5-thinking");
+  assert.equal(block.build.temperature, 0.2);
+  assert.equal(block.plan.model, "omniroute/claude-opus-4-5-thinking");
+  assert.equal(block.plan.top_p, 0.95);
+  assert.equal(block.review.model, "omniroute/gemini-3-flash");
+  assert.equal(block.review.temperature, 0.0);
+});
+
+test("createOmniRouteAgentBlock omits optional fields when not supplied", () => {
+  const block = createOmniRouteAgentBlock({
+    roles: { build: { modelId: "claude-sonnet-4-5-thinking" } },
+  });
+  assert.equal(block.build.model, "omniroute/claude-sonnet-4-5-thinking");
+  assert.ok(!("temperature" in block.build));
+  assert.ok(!("top_p" in block.build));
+  assert.ok(!("tools" in block.build));
+  assert.ok(!("prompt" in block.build));
+});
+
+test("createOmniRouteAgentBlock skips roles with empty modelId", () => {
+  const block = createOmniRouteAgentBlock({
+    roles: {
+      build: { modelId: "claude-sonnet-4-5-thinking" },
+      plan: { modelId: "   " },
+      review: { modelId: "" },
+    },
+  });
+  assert.deepEqual(Object.keys(block), ["build"]);
+});
+
+test("createOmniRouteAgentBlock emits tools as Record<string, boolean> per OC schema", () => {
+  const block = createOmniRouteAgentBlock({
+    roles: {
+      build: {
+        modelId: "claude-sonnet-4-5-thinking",
+        tools: { edit: true, bash: true, web: false },
+        prompt: "Edit files carefully.",
+      },
+    },
+  });
+  assert.deepEqual(block.build.tools, { edit: true, bash: true, web: false });
+  assert.equal(block.build.prompt, "Edit files carefully.");
+});
+
+test("createOmniRouteAgentBlock filters invalid tool entries and omits empty maps", () => {
+  const block = createOmniRouteAgentBlock({
+    roles: {
+      build: {
+        modelId: "claude-sonnet-4-5-thinking",
+        // @ts-expect-error — exercising runtime guard against bad input
+        tools: { edit: true, bash: "yes", "": true, web: null },
+      },
+      plan: {
+        modelId: "claude-opus-4-5-thinking",
+        tools: {},
+      },
+    },
+  });
+  assert.deepEqual(block.build.tools, { edit: true });
+  assert.ok(!("tools" in block.plan));
+});
+
+test("createOmniRouteModesBlock builds provider-prefixed mode entries", () => {
+  const block = createOmniRouteModesBlock({
+    modes: {
+      build: { modelId: "claude-sonnet-4-5-thinking", tools: { edit: true, bash: true } },
+      plan: { modelId: "claude-opus-4-5-thinking", prompt: "Plan first, code later." },
+      review: { modelId: "gemini-3-flash" },
+    },
+  });
+  assert.equal(block.build.model, "omniroute/claude-sonnet-4-5-thinking");
+  assert.deepEqual(block.build.tools, { edit: true, bash: true });
+  assert.equal(block.plan.prompt, "Plan first, code later.");
+  assert.equal(block.review.model, "omniroute/gemini-3-flash");
+});
+
+test("createOmniRouteModesBlock skips modes with empty modelId", () => {
+  const block = createOmniRouteModesBlock({
+    modes: {
+      build: { modelId: "claude-sonnet-4-5-thinking" },
+      plan: { modelId: "" },
+    },
+  });
+  assert.deepEqual(Object.keys(block), ["build"]);
+});
+
+test("createOmniRouteModesBlock honours numeric overrides limited to OC schema", () => {
+  const block = createOmniRouteModesBlock({
+    modes: {
+      build: {
+        modelId: "claude-sonnet-4-5-thinking",
+        temperature: 0.7,
+        top_p: 0.9,
+      },
+    },
+  });
+  assert.equal(block.build.temperature, 0.7);
+  assert.equal(block.build.top_p, 0.9);
+});
+
+// #3419 — soft-deprecation in favour of @omniroute/opencode-plugin. Guard the
+// deprecation notice so it can't be silently dropped while the package is kept
+// publishing (it still works; it is just no longer the recommended path).
+test("package is marked deprecated in favour of @omniroute/opencode-plugin (#3419)", () => {
+  const here = dirname(fileURLToPath(import.meta.url));
+  const pkg = JSON.parse(readFileSync(join(here, "..", "package.json"), "utf8"));
+  assert.match(pkg.description, /DEPRECATED/);
+  assert.match(pkg.description, /@omniroute\/opencode-plugin/);
+
+  const readme = readFileSync(join(here, "..", "README.md"), "utf8");
+  assert.match(readme, /Deprecated/i);
+  assert.match(readme, /@omniroute\/opencode-plugin/);
+});

@omniroute/opencode-provider/tsconfig.json

@@ -0,0 +1,20 @@
+{
+  "compilerOptions": {
+    "target": "ES2022",
+    "module": "ESNext",
+    "moduleResolution": "Bundler",
+    "lib": ["ES2022"],
+    "types": ["node"],
+    "strict": true,
+    "esModuleInterop": true,
+    "skipLibCheck": true,
+    "declaration": true,
+    "isolatedModules": true,
+    "forceConsistentCasingInFileNames": true,
+    "noUncheckedIndexedAccess": false,
+    "outDir": "dist",
+    "rootDir": "src"
+  },
+  "include": ["src/**/*.ts"],
+  "exclude": ["dist", "node_modules", "tests"]
+}

@omniroute/opencode-provider/tsup.config.ts

@@ -0,0 +1,18 @@
+import { defineConfig } from "tsup";
+
+export default defineConfig({
+  entry: ["src/index.ts"],
+  format: ["esm", "cjs"],
+  dts: true,
+  clean: true,
+  sourcemap: false,
+  splitting: false,
+  treeshake: true,
+  target: "node22",
+  outDir: "dist",
+  minify: false,
+});
+
+// CJS consumers should prefer named imports (`require(pkg).createOmniRouteProvider`).
+// The `default` export is also exposed for ESM ergonomics, which makes tsup warn
+// about mixed exports — that's expected and harmless for this package.

AGENTS.md

@@ -3,104 +3,594 @@
 ## Project
 
 Unified AI proxy/router — route any LLM through one endpoint. Multi-provider support
-(OpenAI, Anthropic, Gemini, DeepSeek, Groq, xAI, Mistral, Fireworks, Cohere, etc.)
+with **231 provider entries** (OpenAI, Anthropic, Gemini, DeepSeek, Groq, xAI, Mistral, Fireworks,
+Cohere, NVIDIA, Cerebras, Pollinations, Puter, Cloudflare AI, HuggingFace, DeepInfra,
+SambaNova, Meta Llama API, Moonshot AI, AI21 Labs, Databricks, Snowflake, and many more)
+with **MCP Server** (87 tools), **A2A v0.3 Protocol**, and **Electron desktop app**.
+
+> **Live counts (v3.8.31)**: providers 231 · MCP tools 87 · MCP scopes 30 · A2A skills 6 ·
+> open-sse services 115 · routing strategies 15 · auto-combo scoring factors 9 ·
+> DB modules 83 · DB migrations 97 · base tables 17 · search providers 11 ·
+> i18n locales 42. **Refresh with `npm run check:docs-all`.**
+
+## Doc Accuracy Discipline (read before writing any doc)
+
+> **If `grep -rn "name" src/ open-sse/ bin/` returns nothing, the name does not exist. Do not document it.**
+
+The recurring failure mode in AI-generated docs is _plausible-but-unverified specifics_.
+Every claim in a `.md` file under `docs/` should be verifiable against the source.
+
+**Rules (enforced by `npm run check:fabricated-docs`):**
+
+1. **Never state an API name, endpoint, path, CLI command, or env var without grepping for it first.**
+   ```bash
+   grep -rn "theName" src/ open-sse/ bin/
+   # 0 hits → do not document
+   ```
+2. **Never write a line count, file size, migration count, provider count, or strategy count from memory.**
+   ```bash
+   wc -l <file>           # exact line count
+   ls <dir>/*.ts | wc -l  # file count
+   ```
+3. **Every code example should be copy-pasted from real usage or actually run** — not synthesized.
+   Link to a real call site (`path:line`) instead of inventing a signature.
+4. **Prefer citing real source (`file.ts:line`) over paraphrasing behavior** — verifiable and self-correcting.
+5. **A shorter doc that is 100% accurate beats a comprehensive one with fabrications.**
+   Wrong docs cost more than missing docs, because people trust and act on them.
+
+The script `scripts/check/check-fabricated-docs.mjs` extracts every route path, env var, hook
+name, function name, and file reference from `docs/**/*.md` and verifies each one against the
+codebase. Run it locally before pushing docs; it runs in CI via `npm run check:docs-all`.
 
 ## Stack
 
-- **Runtime**: Next.js 16 (App Router), Node.js, ES Modules
-- **Language**: TypeScript 5.9 (`src/`) + JavaScript (`open-sse/`)
+- **Runtime**: Next.js 16 (App Router), Node.js `>=22.0.0 <23 || >=24.0.0 <27`, ES Modules (`"type": "module"`)
+- **Language**: TypeScript 6.0 (`src/`) + JavaScript (`open-sse/`, `electron/`)
 - **Database**: better-sqlite3 (SQLite) — `DATA_DIR` configurable, default `~/.omniroute/`
-- **Streaming**: SSE via `open-sse` internal package
+- **Streaming**: SSE via `open-sse` internal workspace package
 - **Styling**: Tailwind CSS v4
-- **Docker**: Multi-stage Dockerfile, 3 profiles (base / cli / host)
+- **i18n**: next-intl with 42 locales (`src/i18n/messages/`) — refresh with `ls src/i18n/messages/*.json | wc -l`
+- **Desktop**: Electron (cross-platform: Windows, macOS, Linux)
+- **Schemas**: Zod v4 for all API / MCP input validation
+
+---
+
+## Build, Lint, and Test Commands
+
+| Command                             | Description                                                        |
+| ----------------------------------- | ------------------------------------------------------------------ |
+| `npm run dev`                       | Start Next.js dev server                                           |
+| `npm run build`                     | Production build: `next build` → `.build/next/` + assemble `dist/` |
+| `npm run build:release`             | Clean rebuild + HEAD sentinel (`dist/BUILD_SHA`) — use for deploy  |
+| `npm run start`                     | Run production build                                               |
+| `npm run build:cli`                 | Build CLI package                                                  |
+| `npm run lint`                      | ESLint on all source files                                         |
+| `npm run typecheck:core`            | TypeScript core type checking                                      |
+| `npm run typecheck:noimplicit:core` | Strict checking (no implicit any)                                  |
+| `npm run check`                     | Run lint + test                                                    |
+| `npm run check:cycles`              | Check for circular dependencies                                    |
+| `npm run electron:dev`              | Run Electron app in dev mode                                       |
+| `npm run electron:build`            | Build Electron app for current OS                                  |
+
+**Build output layout:**
+
+| Directory | Purpose                                            | Gitignored |
+| --------- | -------------------------------------------------- | ---------- |
+| `src/`    | Application source (TypeScript / TSX)              | No         |
+| `.build/` | Build intermediates (`distDir = .build/next`)      | Yes        |
+| `dist/`   | Shippable bundle assembled by `assembleStandalone` | Yes        |
+
+The pipeline is a single `next build` pass — intermediates land in `.build/next/`, the
+assembled bundle in `dist/`. VPS deploys rsync `dist/` into the remote
+`/usr/lib/node_modules/omniroute/app/` directory (VPS image path is unchanged).
+
+### Running Tests
+
+```bash
+# All tests (unit + vitest + ecosystem + e2e)
+npm run test:all
+
+# Single test file (Node.js native test runner — most tests use this)
+node --import tsx/esm --test tests/unit/your-file.test.ts
+node --import tsx/esm --test tests/unit/plan3-p0.test.ts
+node --import tsx/esm --test tests/unit/fixes-p1.test.ts
+node --import tsx/esm --test tests/unit/security-fase01.test.ts
+
+# Integration tests
+node --import tsx/esm --test tests/integration/*.test.ts
+
+# Vitest (MCP server, autoCombo)
+npm run test:vitest
+
+# E2E with Playwright
+npm run test:e2e
+
+# Protocol clients E2E (MCP transports, A2A)
+npm run test:protocols:e2e
+
+# Ecosystem compatibility tests
+npm run test:ecosystem
+
+# Coverage (see CONTRIBUTING.md)
+npm run test:coverage
+```
+
+**For authoritative coverage requirements, test execution, and PR gates, see [`CONTRIBUTING.md`](CONTRIBUTING.md#running-tests).**
+
+---
+
+## Code Style Guidelines
+
+### Formatting (Prettier — enforced via lint-staged)
+
+2 spaces · semicolons required · double quotes (`"`) · 100 char width · es5 trailing commas.
+Always run `prettier --write` on changed files.
+
+### TypeScript
+
+- **Target**: ES2022 · **Module**: `esnext` · **Resolution**: `bundler`
+- `strict: false` — prefer explicit types, don't rely on inference
+- Path aliases: `@/*` → `src/`, `@omniroute/open-sse` → `open-sse/`, `@omniroute/open-sse/*` → `open-sse/*`
+
+### ESLint Rules
+
+- **Security (error, everywhere)**: `no-eval`, `no-implied-eval`, `no-new-func`
+- **Relaxed in `open-sse/` and `tests/`**: `@typescript-eslint/no-explicit-any` = warn
+- React hooks rules and `@next/next/no-assign-module-variable` disabled in `open-sse/` and `tests/`
+
+### Naming
+
+| Element             | Convention                       | Example                              |
+| ------------------- | -------------------------------- | ------------------------------------ |
+| Files               | camelCase / kebab-case           | `chatCore.ts`, `tokenHealthCheck.ts` |
+| React components    | PascalCase                       | `Dashboard.tsx`, `ProviderCard.tsx`  |
+| Functions/variables | camelCase                        | `getHealth()`, `switchCombo()`       |
+| Constants           | UPPER_SNAKE                      | `MAX_RETRIES`, `DEFAULT_TIMEOUT`     |
+| Interfaces          | PascalCase (`I` prefix optional) | `ProviderConfig`                     |
+| Enums               | PascalCase (members too)         | `LogLevel.Error`                     |
+
+### Imports
+
+- **Order**: external → internal (`@/`, `@omniroute/open-sse`) → relative (`./`, `../`)
+- **No barrel imports** from `localDb.ts` — import from the specific `db/` module instead
+
+### Error Handling
+
+- try/catch with specific error types; always log with context (pino logger)
+- Never silently swallow errors in SSE streams — use abort signals for cleanup
+- Return proper HTTP status codes (4xx client, 5xx server)
+
+### Security
+
+- **NEVER** commit API keys, secrets, or credentials
+- Validate all user inputs with Zod schemas
+- Auth middleware required on all API routes
+- Never log SQLite encryption keys
+- Sanitize user content (dompurify for HTML)
+- **Public upstream OAuth identifiers** (Gemini / Antigravity / Windsurf-style client_id/secret + Firebase Web keys extracted from public CLIs): use `resolvePublicCred()` from `open-sse/utils/publicCreds.ts`, **never** as string literals. Full pattern in `docs/security/PUBLIC_CREDS.md`.
+- **Error responses** (HTTP / SSE / executor / MCP): use `buildErrorBody()` or `sanitizeErrorMessage()` from `open-sse/utils/error.ts`, **never** put raw `err.stack` / `err.message` in a Response body. Full pattern in `docs/security/ERROR_SANITIZATION.md`.
+- **`exec()` / `spawn()` with runtime values**: pass via the `env` option, **never** string-interpolate paths/values into the script body. Reference: `src/mitm/cert/install.ts::updateNssDatabases`.
+- Prefer secure-by-default libraries when available — see [tldrsec/awesome-secure-defaults](https://github.com/tldrsec/awesome-secure-defaults) for the curated list (Helmet.js, DOMPurify, ssrf-req-filter, safe-regex, Google Tink, etc.).
+
+---
 
 ## Architecture
 
 ### Data Layer (`src/lib/db/`)
 
-All persistence uses SQLite through domain-specific modules:
+All persistence uses SQLite through **83 domain-specific modules** in `src/lib/db/`. Top modules:
 
-| Module         | Responsibility                             |
-| -------------- | ------------------------------------------ |
-| `core.ts`      | SQLite engine, migrations, WAL, encryption |
-| `providers.ts` | Provider connections & nodes               |
-| `models.ts`    | Model aliases, MITM aliases, custom models |
-| `combos.ts`    | Combo configurations                       |
-| `apiKeys.ts`   | API key management & validation            |
-| `settings.ts`  | Settings, pricing, proxy config            |
-| `backup.ts`    | Backup / restore operations                |
+- Core: `core.ts`, `migrationRunner.ts`, `encryption.ts`, `stateReset.ts`
+- Providers / catalog: `providers.ts`, `models.ts`, `providerLimits.ts`, `compressionAnalytics.ts`
+- Routing: `combos.ts`, `modelComboMappings.ts`, `domainState.ts`, `commandCodeAuth.ts`
+- Auth: `apiKeys.ts`, `secrets.ts`, `registeredKeys.ts`, `sessionAccountAffinity.ts`
+- Usage / billing: `quotaSnapshots.ts`, `creditBalance.ts`, `usage*.ts`, `compressionCacheStats.ts`
+- Storage: `backup.ts`, `cleanup.ts`, `jsonMigration.ts`, `healthCheck.ts`, `databaseSettings.ts`
+- Extension modules: `evals.ts`, `webhooks.ts`, `reasoningCache.ts`, `readCache.ts`, `tierConfig.ts`, `compressionCombos.ts`, `compressionScheduler.ts`, `batches.ts`, `files.ts`, `syncTokens.ts`, `proxies.ts`, `oneproxy.ts`, `upstreamProxy.ts`, `versionManager.ts`, `cliToolState.ts`, `prompts.ts`, `detailedLogs.ts`, `contextHandoffs.ts`, `compression.ts`, `stats.ts`
 
-`src/lib/localDb.ts` is a **re-export layer only** — all 27+ consumers import from it,
-but the real logic lives in `src/lib/db/`.
+Live count: `ls src/lib/db/*.ts | wc -l` (currently 83). Drift detection: `npm run check:docs-counts`.
+Schema migrations live in `db/migrations/` (**97 files** as of v3.8.24) and run via `migrationRunner.ts`.
+`src/lib/localDb.ts` is a **re-export layer only** — never add logic there.
+
+#### DB Internals
+
+- **`core.ts`**: `getDbInstance()` returns a singleton `better-sqlite3` instance with WAL
+  journaling. `SCHEMA_SQL` defines **17 base tables** (verify with `grep -c "CREATE TABLE" src/lib/db/core.ts` minus 1 for the bookkeeping `_omniroute_migrations` table). Helpers: `rowToCamel`, `encryptConnectionFields`.
+- **`migrationRunner.ts`**: Applies versioned SQL files from `db/migrations/` inside transactions.
+  Tracks applied migrations in `_omniroute_migrations` table.
+- **Migrations**: 97 files (`001_initial_schema.sql` → `099_*.sql`).
+  Each migration is idempotent and runs in a transaction. Live count: `ls src/lib/db/migrations/*.sql | wc -l`.
+- **Domain modules** import `getDbInstance()` from `core.ts` for all CRUD operations.
+  Each module owns a specific table/set of tables (e.g., `providers.ts` → `provider_connections`,
+  `combos.ts` → `combos`). Encryption helpers protect sensitive fields at rest.
+- **`localDb.ts`** re-exports all domain modules — consumers import from here for convenience.
+
+### API Route Layer (`src/app/api/v1/`)
+
+Next.js App Router routes — each follows a consistent pattern:
+
+```
+Route → CORS preflight → Body validation (Zod) → Optional auth (extractApiKey/isValidApiKey)
+  → API key policy enforcement (enforceApiKeyPolicy) → Handler delegation (open-sse)
+```
+
+| Route                           | Handler                   | Notes                                                         |
+| ------------------------------- | ------------------------- | ------------------------------------------------------------- |
+| `chat/completions/route.ts`     | `handleChat()`            | + prompt injection guard (clones request)                     |
+| `responses/route.ts`            | `handleChat()` (unified)  | Responses API format                                          |
+| `embeddings/route.ts`           | `handleEmbedding()`       | Model listing + creation                                      |
+| `images/generations/route.ts`   | `handleImageGeneration()` | Model listing + creation                                      |
+| `audio/transcriptions/route.ts` | audio handler             | Multipart form data                                           |
+| `audio/speech/route.ts`         | TTS handler               | Binary audio response                                         |
+| `videos/generations/route.ts`   | video handler             | ComfyUI/SD WebUI                                              |
+| `music/generations/route.ts`    | music handler             | ComfyUI workflows                                             |
+| `moderations/route.ts`          | moderation handler        | Content safety                                                |
+| `rerank/route.ts`               | rerank handler            | Document relevance                                            |
+| `search/route.ts`               | search handler            | Web search (12 providers per `open-sse/handlers/search.ts:6`) |
+
+**No global Next.js middleware file** — interception is route-specific. Auth is optional
+(controlled by `REQUIRE_API_KEY` env). Prompt injection guard is unique to chat completions.
 
 ### Request Pipeline (`open-sse/`)
 
-| Handler                 | Role                                        |
-| ----------------------- | ------------------------------------------- |
-| `chatCore.js`           | Main chat completions proxy (SSE / non-SSE) |
-| `responsesHandler.js`   | OpenAI Responses API compat                 |
-| `responseTranslator.js` | Format translation for Responses API        |
-| `embeddings.js`         | Embedding proxy                             |
-| `imageGeneration.js`    | Image generation proxy                      |
-| `sseParser.js`          | SSE stream parser                           |
-| `usageExtractor.js`     | Token usage extraction from responses       |
+The `open-sse/` workspace is the core streaming engine. Full request flow:
 
-Translation between provider formats: `open-sse/translator/`
+```
+Client Request
+  → src/app/api/v1/.../route.ts (Next.js route)
+    → open-sse/handlers/chatCore.ts::handleChatCore()
+      → Semantic/signature cache check
+      → Rate limit check (rateLimitManager)
+      → Combo routing? → open-sse/services/combo.ts::handleComboChat()
+        → resolveComboTargets() → ordered ResolvedComboTarget[]
+        → For each target: handleSingleModel() (wraps chatCore)
+      → translateRequest() (open-sse/translator/)
+        → Convert source format (e.g., OpenAI) → target format (e.g., Claude)
+      → getExecutor() → provider-specific executor instance
+        → executor.execute() (BaseExecutor → DefaultExecutor or provider-specific)
+          → buildUrl() + buildHeaders() + transformRequest()
+          → fetch() to upstream provider
+          → Retry logic with exponential backoff
+      → Response translation back to client format
+      → If Responses API: responsesTransformer.ts TransformStream
+  → SSE stream or JSON response to client
+```
 
-### OAuth & Tokens (`src/lib/oauth/`)
+**Handlers** (`open-sse/handlers/`): `chatCore.ts`, `responsesHandler.ts`, `embeddings.ts`,
+`imageGeneration.ts`, `videoGeneration.ts`, `musicGeneration.ts`, `audioSpeech.ts`,
+`audioTranscription.ts`, `moderations.ts`, `rerank.ts`, `search.ts`.
 
-18 modules handling OAuth flows, token refresh, and provider credentials.
-Default credentials are hardcoded in `src/lib/oauth/constants/oauth.ts`,
-overridable via env vars or `data/provider-credentials.json`.
+**Upstream headers**: merged after default auth; same header name replaces executor value.
+**T5 intra-family fallback** recomputes headers using only the fallback model id.
+Forbidden header names: `src/shared/constants/upstreamHeaders.ts` — keep sanitize,
+Zod schemas, and unit tests aligned when editing.
 
-### Supporting Systems
+### Provider Categories
 
-| System                     | Location                                          |
-| -------------------------- | ------------------------------------------------- |
-| Usage tracking & analytics | `src/lib/usageDb.ts`, `src/lib/usageAnalytics.ts` |
-| Token health checks        | `src/lib/tokenHealthCheck.ts`                     |
-| Cloud sync                 | `src/lib/cloudSync.ts`                            |
-| Proxy logging              | `src/lib/proxyLogger.ts`                          |
-| Data paths resolution      | `src/lib/dataPaths.ts`                            |
+- **Free** (4): Qoder AI, Qwen Code, Gemini CLI (deprecated), Kiro AI
+- **OAuth** (14): Claude Code, Antigravity, Codex, GitHub Copilot, Cursor, Kimi Coding, Kilo Code, Cline, Qwen (⚠️ free tier discontinued 2026-04-15), Kiro, Qoder, Gemini, Windsurf (v3.8), GitLab Duo (v3.8)
+- **API Key** (120+): OpenAI, Anthropic, Gemini, DeepSeek, Groq, xAI, Mistral, Perplexity,
+  Together, Fireworks, Cerebras, Cohere, NVIDIA, Nebius, SiliconFlow, Hyperbolic,
+  HuggingFace, OpenRouter, Vertex AI, Cloudflare AI, Scaleway, AI/ML API, Pollinations,
+  Puter, Longcat, Alibaba, Kimi, Minimax, Blackbox, Synthetic, Kilo Gateway,
+  Z.AI, GLM, Deepgram, AssemblyAI, ElevenLabs, Cartesia, PlayHT, Inworld,
+  NanoBanana, SD WebUI, ComfyUI, Ollama Cloud, Perplexity Search, Serper, Brave, Exa,
+  Tavily, OpenCode Zen/Go, Bailian Coding Plan, DeepInfra, Vercel AI Gateway,
+  Lambda AI, SambaNova, nScale, OVHcloud AI, Baseten, PublicAI, Moonshot AI,
+  Meta Llama API, v0 (Vercel), Morph, Featherless AI, FriendliAI, LlamaGate,
+  Galadriel, Weights & Biases Inference, Volcengine, AI21 Labs, Venice.ai,
+  Codestral, Upstage, Maritalk, Xiaomi MiMo, Inference.net, NanoGPT, Predibase,
+  Bytez, Heroku AI, Databricks, Snowflake Cortex, GigaChat (Sber), CrofAI,
+  AgentRouter, ChatGPT Web, Baidu Qianfan, AWS Polly, RunwayML, GitLab Duo,
+  Amazon Q, Empower, Poe, and many more.
+- **Self-Hosted** (8+): LM Studio, vLLM, Lemonade, Llamafile, Triton, Docker Model Runner, Xinference, Oobabooga
+- **Custom**: OpenAI-compatible (`openai-compatible-*`) and Anthropic-compatible (`anthropic-compatible-*`) prefixes
+
+Providers are registered in `src/shared/constants/providers.ts` with Zod validation at module load.
+
+### Executors (`open-sse/executors/`)
+
+Provider-specific request executors: `base.ts`, `default.ts`, `cursor.ts`, `codex.ts`,
+`antigravity.ts`, `github.ts`, `gemini-cli.ts`, `kiro.ts`, `qoder.ts`, `vertex.ts`,
+`cloudflare-ai.ts`, `opencode.ts`, `pollinations.ts`, `puter.ts`.
+
+#### Executor Internals
+
+- **`base.ts`** (`BaseExecutor`): Abstract base with `buildUrl()`, `buildHeaders()`,
+  `transformRequest()`, retry logic (exponential backoff), and `execute()`. Subclasses
+  override URL/header/transform methods for provider-specific behavior.
+- **`default.ts`** (`DefaultExecutor extends BaseExecutor`): Handles most OpenAI-compatible
+  providers. Reads provider config from `providerRegistry.ts` to resolve base URL, auth
+  header format, and request transformations.
+- **`getExecutor()`** (`executors/index.ts`): Factory that returns the correct executor
+  instance based on provider ID. Provider-specific executors (Cursor, Codex, Vertex, etc.)
+  override only what differs from the default.
+
+### Translator (`open-sse/translator/`)
+
+Translates between API formats (OpenAI-format ↔ Anthropic, Gemini, etc.).
+Includes request/response translators with helpers for image handling.
+
+#### Translator Internals
+
+- **`translator/index.ts`**: Exports `translateRequest()` and format constants. Called by
+  `chatCore.ts` before executor dispatch.
+- **Flow**: `translateRequest(body, sourceFormat, targetFormat)` → detects source format
+  (OpenAI, Anthropic, Gemini) → applies the matching translator module → returns
+  transformed body ready for the target provider.
+- **Response translation** runs in reverse after upstream response, converting back to
+  the client's expected format.
+
+### Transformer (`open-sse/transformer/`)
+
+`responsesTransformer.ts` — transforms Responses API format to/from Chat Completions format.
+
+#### Transformer Internals
+
+- **`createResponsesApiTransformStream()`**: Returns a `TransformStream` that converts
+  Chat Completions SSE chunks (`data: {"choices":[...]}`) into Responses API SSE events
+  (`response.output_item.added`, `response.output_text.delta`, etc.).
+- Used when the client sends a Responses API request: the request is internally converted
+  to Chat Completions format, dispatched normally, and the response is piped through this
+  transform stream before reaching the client.
+
+### Services (`open-sse/services/`)
+
+115 service modules in `open-sse/services/` (top-level only; 184 including sub-dirs like `autoCombo/` and `compression/`). Refresh: `ls open-sse/services/*.ts | wc -l`. Key modules:
+`combo.ts` (routing engine), `usage.ts`, `tokenRefresh.ts`,
+`rateLimitManager.ts`, `accountFallback.ts`, `sessionManager.ts`, `wildcardRouter.ts`,
+`autoCombo/`, `intentClassifier.ts`, `taskAwareRouter.ts`, `thinkingBudget.ts`,
+`contextManager.ts`, `modelDeprecation.ts`, `modelFamilyFallback.ts`,
+`emergencyFallback.ts`, `workflowFSM.ts`, `backgroundTaskDetector.ts`, `ipFilter.ts`,
+`signatureCache.ts`, `volumeDetector.ts`, `contextHandoff.ts`, `compression/` (prompt
+compression pipeline), and more.
+
+#### Prompt Compression Pipeline (`compression/`)
+
+Modular prompt compression that runs proactively before the existing reactive context manager.
+
+- **`strategySelector.ts`**: Selects compression mode based on config, compression combo assignments,
+  combo overrides, auto-trigger thresholds, and defaults. Priority: assigned compression combo >
+  combo override > auto-trigger > default mode > off.
+- **`lite.ts`**: 5 lite-mode techniques: `collapseWhitespace`, `dedupSystemPrompt`,
+  `compressToolResults`, `removeRedundantContent`, `replaceImageUrls`. Target: 10-15% savings at
+  <1ms latency.
+- **`caveman.ts` / `cavemanRules.ts`**: Caveman-style semantic condensation backed by built-in
+  rules plus file-loaded language packs under `compression/rules/`.
+- **`engines/rtk/`**: Rule-based terminal/tool-output compression inspired by RTK patterns. Detects
+  command output classes, applies JSON filter packs, deduplicates repeated lines, strips ANSI/code
+  noise, and preserves errors/actionable context. The RTK JSON DSL supports replace,
+  match-output short-circuit, strip/keep, per-line truncation, head/tail/max-line truncation,
+  inline tests, trust-gated project/global custom filters, and optional redacted raw-output
+  retention for authenticated recovery.
+- **`engines/registry.ts`**: Registers engines (`caveman`, `rtk`) and powers stacked pipelines.
+- **`stats.ts`**: Per-request compression stats tracking (original tokens, compressed tokens,
+  savings %, techniques used, engine breakdown, compression combo id).
+- **`types.ts`**: `CompressionMode` (off/lite/standard/aggressive/ultra/rtk/stacked),
+  `CompressionConfig`, `CompressionStats`, `CompressionResult`.
+- DB settings in `src/lib/db/compression.ts`, compression combos in
+  `src/lib/db/compressionCombos.ts`, API routes under `src/app/api/settings/compression/`,
+  `src/app/api/context/*`, and preview/language-pack routes under `src/app/api/compression/*`.
+
+#### Combo Routing Engine (`combo.ts`)
+
+- **`handleComboChat()`**: Entry point for combo-routed requests. Receives the combo config
+  and iterates through targets in order until one succeeds or all fail.
+- **`resolveComboTargets()`**: Expands a combo configuration into an ordered array of
+  `ResolvedComboTarget[]`, each specifying provider + model + account + credentials.
+- **Strategies** (15): priority, weighted, fill-first, round-robin, P2C, random, least-used, reset-aware (v3.8),
+  reset-window, cost-optimized, strict-random, auto, lkgp, context-optimized, context-relay. Source: `ROUTING_STRATEGY_VALUES` in `src/shared/constants/routingStrategies.ts`.
+- Each target calls **`handleSingleModel()`** which wraps `handleChatCore()` with
+  per-target error handling and circuit breaker checks.
+
+### Domain Layer (`src/domain/`)
+
+Policy engine modules: `policyEngine.ts`, `comboResolver.ts`, `costRules.ts`,
+`degradation.ts`, `fallbackPolicy.ts`, `lockoutPolicy.ts`, `modelAvailability.ts`,
+`providerExpiration.ts`, `quotaCache.ts`, `responses.ts`, `configAudit.ts`.
+
+### MCP Server (`open-sse/mcp-server/`)
+
+**87 tools** total (`TOTAL_MCP_TOOL_COUNT`, `open-sse/mcp-server/server.ts`): a 33-entry base registry (`MCP_TOOLS` in `schemas/tools.ts`, bundling the core / cache / compression / 1proxy / advanced tools) **plus** standalone module sets — memory (3), skill (4), agentSkill (3), gamification (8), plugin (8), notion (6), obsidian (22). 3 transports (stdio / SSE / Streamable HTTP). Scoped auth (30 scopes — see `OMNIROUTE_MCP_SCOPES`), Zod schemas. See [`docs/frameworks/MCP-SERVER.md`](docs/frameworks/MCP-SERVER.md).
+
+**Core tools** (20): get_health, list_combos, get_combo_metrics, switch_combo, check_quota,
+route_request, cost_report, list_models_catalog, web_search, simulate_route, set_budget_guard,
+set_routing_strategy, set_resilience_profile, test_combo, get_provider_metrics,
+best_combo_for_task, explain_route, get_session_snapshot, db_health_check, sync_pricing.
+
+**Cache tools** (2): cache_stats, cache_flush.
+
+**Compression tools** (5): compression_status, compression_configure, set_compression_engine,
+list_compression_combos, compression_combo_stats.
+
+**1proxy tools** (3): oneproxy_fetch, oneproxy_rotate, oneproxy_stats.
+
+**Memory tools** (3): memory_search, memory_add, memory_clear.
+
+**Skill tools** (4): skills_list, skills_enable, skills_execute, skills_executions.
+
+**Agent-skill tools** (3): A2A skill discovery / invocation bridges.
+
+**Gamification tools** (8): levels, badges, leaderboard, and community-federation queries.
+
+**Plugin tools** (8): plugin marketplace listing, install/enable/disable, and runtime inspection.
+
+**Notion tools** (6) + **Obsidian tools** (22): knowledge-base read/write integrations (the largest tool family — vault search, note CRUD, WebDAV-backed file ops).
+
+#### MCP Internals
+
+- **Tool registration**: Each tool is an object with `{ name, description, inputSchema: ZodSchema,
+handler: async (args) => {...} }`. Zod validates inputs before the handler fires.
+- **`createMcpServer()`** and **`startMcpStdio()`** exported from `mcp-server/index.ts`.
+  `createMcpServer()` wires all tool sets; `startMcpStdio()` launches the stdio transport.
+- **Transports**: stdio (CLI `omniroute --mcp`), SSE (`/api/mcp/sse`), Streamable HTTP
+  (`/api/mcp/stream`). All share the same tool/scope engine.
+- **Scopes** (30): Control which tool categories an API key can access. Enforcement happens
+  before handler dispatch.
+- **Audit**: Every tool invocation is logged to SQLite (`mcp_audit` table) with tool name,
+  args, success/failure, API key attribution, and timestamp.
+
+### A2A Server (`src/lib/a2a/`)
+
+JSON-RPC 2.0, SSE streaming, Task Manager with TTL cleanup.
+Agent Card at `/.well-known/agent.json`.
+Skills (6): `smartRouting.ts`, `quotaManagement.ts`, `providerDiscovery.ts`, `costAnalysis.ts`, `healthReport.ts`, `listCapabilities.ts`.
+
+#### A2A Internals
+
+- **`taskManager.ts`**: State machine lifecycle for tasks: `submitted → working →
+completed | failed | canceled`. Tasks have TTL and are cleaned up automatically.
+- **JSON-RPC methods**: `message/send` (sync), `message/stream` (SSE), `tasks/get`,
+  `tasks/cancel`. Dispatched via `POST /a2a`.
+- **Skills**: Registered in a DB-backed registry. Each skill receives task context
+  (messages, metadata) and returns structured results. `quotaManagement.ts` summarizes
+  quota; `smartRouting.ts` recommends routing decisions.
+- **Agent Card**: `/.well-known/agent.json` exposes capabilities, skills, and metadata
+  for client auto-discovery.
+
+### ACP Module (`src/lib/acp/`)
+
+Agent Communication Protocol registry and manager.
+
+### Memory System (`src/lib/memory/`)
+
+Extraction, injection, retrieval, summarization, and store modules for persistent
+conversational memory across sessions.
+
+### Skills System (`src/lib/skills/`)
+
+Extensible skill framework: registry, executor, sandbox, built-in skills,
+custom skill support, interception, and injection.
+
+#### Skills Internals
+
+- **`registry.ts`**: DB-backed skill registration and discovery. Skills have metadata
+  (name, description, version, enabled status) stored in SQLite.
+- **`executor.ts`**: Execution engine with configurable timeout and retry logic.
+  Receives skill name + input, looks up the skill, runs it in the sandbox.
+- **`sandbox.ts`**: Isolation layer for custom (user-provided) skills. Limits resource
+  access and execution time.
+- **Built-in skills**: Ship with OmniRoute (e.g., quota management, routing). Located
+  alongside the registry.
+- **Interception/Injection**: Skills can intercept requests in the pipeline (pre/post
+  processing) or inject context into prompts.
+
+### Compliance (`src/lib/compliance/`)
+
+Policy index for compliance enforcement.
+
+### MITM Proxy (`src/mitm/`)
+
+MITM proxy capability with certificate management, DNS handling, and target routing.
+
+### Middleware (`src/middleware/`)
+
+Request middleware including `promptInjectionGuard.ts`.
+
+### Guardrails (`src/lib/guardrails/`)
+
+Hot-reloadable guardrails framework (3 built-in: pii-masker, prompt-injection, vision-bridge). Fail-open; per-request opt-out via header. See [`docs/security/GUARDRAILS.md`](docs/security/GUARDRAILS.md).
+
+### Cloud Agents (`src/lib/cloudAgent/`)
+
+`CloudAgentBase` abstract class + 3 agents (codex-cloud, devin, jules). Tasks persisted in `cloud_agent_tasks`; management auth required. See [`docs/frameworks/CLOUD_AGENT.md`](docs/frameworks/CLOUD_AGENT.md).
+
+### Evals (`src/lib/evals/`)
+
+Generic eval framework: `evalRunner.ts`, `runtime.ts`. Targets: combo / model / suite-default. See [`docs/frameworks/EVALS.md`](docs/frameworks/EVALS.md).
+
+### Webhooks (`src/lib/webhookDispatcher.ts`)
+
+HMAC-signed delivery, exponential backoff, auto-disable after 10 failures. 7 event types. See [`docs/frameworks/WEBHOOKS.md`](docs/frameworks/WEBHOOKS.md).
+
+### Authorization Pipeline (`src/server/authz/`)
+
+`classify → policies → enforce`. 3 route classes (PUBLIC / CLIENT_API / MANAGEMENT). See [`docs/architecture/AUTHZ_GUIDE.md`](docs/architecture/AUTHZ_GUIDE.md).
+
+### Reasoning Replay (`src/lib/db/reasoningCache.ts` + `open-sse/services/reasoningCache.ts`)
+
+Hybrid in-memory + SQLite cache for `reasoning_content`. Re-injects on multi-turn for strict providers (DeepSeek V4, Kimi K2, Qwen-Thinking, GLM, xiaomi-mimo). See [`docs/routing/REASONING_REPLAY.md`](docs/routing/REASONING_REPLAY.md).
+
+### Tunnels (`src/lib/{cloudflaredTunnel,ngrokTunnel}.ts` + `src/app/api/tunnels/`)
+
+Cloudflare Quick/Named, ngrok, Tailscale Funnel. See [`docs/ops/TUNNELS_GUIDE.md`](docs/ops/TUNNELS_GUIDE.md).
 
 ### Adding a New Provider
 
 1. Register in `src/shared/constants/providers.ts`
-2. Add executor in `open-sse/executors/`
-3. Add translator rules in `open-sse/translator/` (if non-OpenAI format)
+2. Add executor in `open-sse/executors/` (if custom logic needed)
+3. Add translator in `open-sse/translator/` (if non-OpenAI format)
 4. Add OAuth config in `src/lib/oauth/constants/oauth.ts` (if OAuth-based)
+5. Add models in `open-sse/config/providerRegistry.ts`
+
+---
+
+## Subdirectory AGENTS.md Files
+
+- **[`src/lib/db/AGENTS.md`](src/lib/db/AGENTS.md)** — SQLite persistence, domain modules, migrations
+- **[`open-sse/services/AGENTS.md`](open-sse/services/AGENTS.md)** — Routing engine, combo resolution, strategy selection
+
+## Reference Documentation (docs/)
+
+For any non-trivial change, read the matching deep-dive first:
+
+| Area                                       | Doc                                                                                                                                 |
+| ------------------------------------------ | ----------------------------------------------------------------------------------------------------------------------------------- |
+| Repo navigation                            | [`docs/architecture/REPOSITORY_MAP.md`](docs/architecture/REPOSITORY_MAP.md)                                                        |
+| Architecture                               | [`docs/architecture/ARCHITECTURE.md`](docs/architecture/ARCHITECTURE.md)                                                            |
+| Engineering reference                      | [`docs/architecture/CODEBASE_DOCUMENTATION.md`](docs/architecture/CODEBASE_DOCUMENTATION.md)                                        |
+| Auto-Combo (12-factor, 15 strategies)      | [`docs/routing/AUTO-COMBO.md`](docs/routing/AUTO-COMBO.md)                                                                          |
+| Resilience (3 layers)                      | [`docs/architecture/RESILIENCE_GUIDE.md`](docs/architecture/RESILIENCE_GUIDE.md)                                                    |
+| Skills                                     | [`docs/frameworks/SKILLS.md`](docs/frameworks/SKILLS.md)                                                                            |
+| Memory                                     | [`docs/frameworks/MEMORY.md`](docs/frameworks/MEMORY.md)                                                                            |
+| Cloud agents                               | [`docs/frameworks/CLOUD_AGENT.md`](docs/frameworks/CLOUD_AGENT.md)                                                                  |
+| Guardrails                                 | [`docs/security/GUARDRAILS.md`](docs/security/GUARDRAILS.md)                                                                        |
+| Evals                                      | [`docs/frameworks/EVALS.md`](docs/frameworks/EVALS.md)                                                                              |
+| Compliance                                 | [`docs/security/COMPLIANCE.md`](docs/security/COMPLIANCE.md)                                                                        |
+| Webhooks                                   | [`docs/frameworks/WEBHOOKS.md`](docs/frameworks/WEBHOOKS.md)                                                                        |
+| Authz                                      | [`docs/architecture/AUTHZ_GUIDE.md`](docs/architecture/AUTHZ_GUIDE.md)                                                              |
+| Stealth                                    | [`docs/security/STEALTH_GUIDE.md`](docs/security/STEALTH_GUIDE.md)                                                                  |
+| Reasoning replay                           | [`docs/routing/REASONING_REPLAY.md`](docs/routing/REASONING_REPLAY.md)                                                              |
+| Agent protocols (A2A / ACP / Cloud)        | [`docs/frameworks/AGENT_PROTOCOLS_GUIDE.md`](docs/frameworks/AGENT_PROTOCOLS_GUIDE.md)                                              |
+| MCP server                                 | [`docs/frameworks/MCP-SERVER.md`](docs/frameworks/MCP-SERVER.md)                                                                    |
+| A2A server                                 | [`docs/frameworks/A2A-SERVER.md`](docs/frameworks/A2A-SERVER.md)                                                                    |
+| API reference                              | [`docs/reference/API_REFERENCE.md`](docs/reference/API_REFERENCE.md) + [`docs/openapi.yaml`](docs/openapi.yaml) |
+| Provider catalog (auto-generated)          | [`docs/reference/PROVIDER_REFERENCE.md`](docs/reference/PROVIDER_REFERENCE.md)                                                      |
+| Tunnels                                    | [`docs/ops/TUNNELS_GUIDE.md`](docs/ops/TUNNELS_GUIDE.md)                                                                            |
+| Electron desktop                           | [`docs/guides/ELECTRON_GUIDE.md`](docs/guides/ELECTRON_GUIDE.md)                                                                    |
+| Release flow                               | [`docs/ops/RELEASE_CHECKLIST.md`](docs/ops/RELEASE_CHECKLIST.md)                                                                    |
+| Quality gates (35 gates, allowlist policy) | [`docs/architecture/QUALITY_GATES.md`](docs/architecture/QUALITY_GATES.md)                                                          |
+| Cluster opt-in profiles (memory, bifrost)  | [`docs/architecture/cluster-decisions.md`](docs/architecture/cluster-decisions.md)                                                  |
+
+---
+
+## Fork / Upstream Workflow
+
+This repository is a fork of `diegosouzapw/OmniRoute`. Keep fork-only operational
+changes (for example GHCR image publishing, personal deployment workflows, or local
+automation) out of upstream contribution PRs.
+
+When preparing a PR for upstream, always start the work branch from `upstream/main`,
+not from this fork's `main`:
+
+```bash
+git fetch upstream
+git switch -c <branch-name> upstream/main
+```
+
+Only cherry-pick or reapply the changes intended for the upstream PR.
+
+---
 
 ## Review Focus
 
-### Security
-
-- No hardcoded API keys or secrets in commits
-- Auth middleware on all API routes
-- Input validation on user-facing endpoints
-- SQLite encryption key must not be logged
-
-### Architecture
-
-- DB operations go through `src/lib/db/` modules, never raw SQL in routes
-- Provider requests flow through `open-sse/handlers/`
-- Translations use `open-sse/translator/` modules
-- `localDb.ts` is re-exports only — add new functions to the proper `db/*.ts` module
-
-### Code Quality
-
-- Consistent error handling with try/catch
-- Proper HTTP status codes
-- No memory leaks in SSE streams (abort signals, cleanup)
-- Rate limit headers must be parsed correctly
-
-### Docker
-
-- Dockerfile has two targets: `runner-base` and `runner-cli`
-- `docker-compose.yml` — development (3 profiles)
-- `docker-compose.prod.yml` — isolated production instance (port 20130)
-- Data persists in named volumes (`omniroute-data` / `omniroute-prod-data`)
-
-### Review Mode
-
-- Provide analysis and suggestions only
-- Focus on bugs, security, performance, and best practices
+- **DB ops** go through `src/lib/db/` modules, never raw SQL in routes
+- **Provider requests** flow through `open-sse/handlers/`
+- **MCP/A2A pages** are tabs inside `/dashboard/endpoint`, not standalone routes
+- **No memory leaks** in SSE streams (abort signals, cleanup)
+- **Rate limit headers** must be parsed correctly
+- All API inputs validated with **Zod schemas**
+- **Provider constants** validated at module load via Zod (`src/shared/validation/providerSchema.ts`)
+- **Pricing data** syncs from LiteLLM via `src/lib/pricingSync.ts`
+- **Memory/Skills** are cross-cutting: affect MCP tools, request pipeline, and A2A skills
+- **⛔ NEVER close a contributor's PR** after using their code — always merge via GitHub so they get credit. See `.agents/workflows/review-prs.md` for full policy.

CLAUDE.md

@@ -0,0 +1,523 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Quick Start
+
+```bash
+npm install                    # Install deps (auto-generates .env from .env.example)
+npm run dev                    # Dev server at http://localhost:20128
+npm run build                  # Production build (Next.js 16 standalone)
+npm run lint                   # ESLint (0 errors expected; warnings are pre-existing)
+npm run typecheck:core         # TypeScript check (should be clean)
+npm run typecheck:noimplicit:core  # Strict check (no implicit any)
+npm run test:coverage          # Unit tests + coverage gate (60/60/60/60 — statements/lines/functions/branches)
+npm run check                  # lint + test combined
+npm run check:cycles           # Detect circular dependencies
+```
+
+### Running Tests
+
+```bash
+# Single test file (Node.js native test runner — most tests)
+node --import tsx/esm --test tests/unit/your-file.test.ts
+
+# Vitest (MCP server, autoCombo, cache)
+npm run test:vitest
+
+# All suites
+npm run test:all
+```
+
+For full test matrix, see `CONTRIBUTING.md` → "Running Tests". For deep architecture, see `AGENTS.md`.
+
+---
+
+## Project at a Glance
+
+**OmniRoute** — unified AI proxy/router. One endpoint, 231 LLM providers, auto-fallback.
+
+| Layer         | Location                | Purpose                                                                                                                                |
+| ------------- | ----------------------- | -------------------------------------------------------------------------------------------------------------------------------------- |
+| API Routes    | `src/app/api/v1/`       | Next.js App Router — entry points                                                                                                      |
+| Handlers      | `open-sse/handlers/`    | Request processing (chat, embeddings, etc)                                                                                             |
+| Executors     | `open-sse/executors/`   | Provider-specific HTTP dispatch                                                                                                        |
+| Translators   | `open-sse/translator/`  | Format conversion (OpenAI↔Claude↔Gemini)                                                                                               |
+| Transformer   | `open-sse/transformer/` | Responses API ↔ Chat Completions                                                                                                       |
+| Services      | `open-sse/services/`    | Combo routing, rate limits, caching, etc                                                                                               |
+| Database      | `src/lib/db/`           | SQLite domain modules (83 files, 97 migrations)                                                                                        |
+| Domain/Policy | `src/domain/`           | Policy engine, cost rules, fallback logic                                                                                              |
+| MCP Server    | `open-sse/mcp-server/`  | 87 tools (33 base + memory/skill/notion/obsidian/gamification/plugin modules), 3 transports (stdio / SSE / Streamable HTTP), 30 scopes |
+| A2A Server    | `src/lib/a2a/`          | JSON-RPC 2.0 agent protocol                                                                                                            |
+| Skills        | `src/lib/skills/`       | Extensible skill framework                                                                                                             |
+| Memory        | `src/lib/memory/`       | Persistent conversational memory                                                                                                       |
+
+Monorepo: `src/` (Next.js 16 app), `open-sse/` (streaming engine workspace), `electron/` (desktop app), `tests/`, `bin/` (CLI entry point).
+
+---
+
+## Request Pipeline
+
+```
+Client → /v1/chat/completions (Next.js route)
+  → CORS → Zod validation → auth? → policy check → prompt injection guard
+  → handleChatCore() [open-sse/handlers/chatCore.ts]
+    → cache check → rate limit → combo routing?
+      → resolveComboTargets() → handleSingleModel() per target
+    → translateRequest() → getExecutor() → executor.execute()
+      → fetch() upstream → retry w/ backoff
+    → response translation → SSE stream or JSON
+    → If Responses API: responsesTransformer.ts TransformStream
+```
+
+API routes follow a consistent pattern: `Route → CORS preflight → Zod body validation → Optional auth (extractApiKey/isValidApiKey) → API key policy enforcement → Handler delegation (open-sse)`. No global Next.js middleware — interception is route-specific.
+
+**Combo routing** (`open-sse/services/combo.ts`): 15 strategies (priority, weighted, fill-first, round-robin, P2C, random, least-used, cost-optimized, reset-aware, reset-window, strict-random, auto, lkgp, context-optimized, context-relay). Each target calls `handleSingleModel()` which wraps `handleChatCore()` with per-target error handling and circuit breaker checks. See `docs/routing/AUTO-COMBO.md` for the 9-factor Auto-Combo scoring and `docs/architecture/RESILIENCE_GUIDE.md` for the 3 resilience layers.
+
+---
+
+## Resilience Runtime State
+
+OmniRoute has three related but distinct temporary-failure mechanisms. Keep their
+scope separate when debugging routing behavior. See the
+[3-layer resilience diagram](./docs/diagrams/exported/resilience-3layers.svg)
+(source: [docs/diagrams/resilience-3layers.mmd](./docs/diagrams/resilience-3layers.mmd))
+for an at-a-glance map.
+
+### Provider Circuit Breaker
+
+**Scope**: whole provider, e.g. `glm`, `openai`, `anthropic`.
+
+**Purpose**: stop sending traffic to a provider that is repeatedly failing at the
+upstream/service level, so one unhealthy provider does not slow down every request.
+
+**Implementation**:
+
+- Core class: `src/shared/utils/circuitBreaker.ts`
+- Chat gate/execution wiring: `src/sse/handlers/chatHelpers.ts`, `src/sse/handlers/chat.ts`
+- Runtime status API: `src/app/api/monitoring/health/route.ts`
+- Shared wrappers: `open-sse/services/accountFallback.ts`
+- Persisted state table: `domain_circuit_breakers`
+
+**States**:
+
+- `CLOSED`: normal traffic is allowed.
+- `OPEN`: provider is temporarily blocked; callers get a provider-circuit-open response
+  or combo routing skips to another target.
+- `HALF_OPEN`: reset timeout has elapsed; allow a probe request. Success closes the
+  breaker, failure opens it again.
+
+**Defaults** (`open-sse/config/constants.ts`):
+
+- OAuth providers: threshold `3`, reset timeout `60s`.
+- API-key providers: threshold `5`, reset timeout `30s`.
+- Local providers: threshold `2`, reset timeout `15s`.
+
+Only provider-level failure statuses should trip the provider breaker:
+
+```ts
+(408, 500, 502, 503, 504);
+```
+
+Do not trip the whole-provider breaker for normal account/key/model errors like most
+`401`, `403`, or `429` cases. Those usually belong to connection cooldown or model
+lockout. A generic API-key provider `403` should be recoverable unless it is classified
+as a terminal provider/account error.
+
+The breaker uses lazy recovery, not a background timer. When `OPEN` expires, reads such
+as `getStatus()`, `canExecute()`, and `getRetryAfterMs()` refresh the state to
+`HALF_OPEN`, so dashboards and combo candidate builders do not keep excluding an
+expired provider forever.
+
+### Connection Cooldown
+
+**Scope**: one provider connection/account/key.
+
+**Purpose**: temporarily skip one bad key/account while allowing other connections for
+the same provider to continue serving requests.
+
+**Implementation**:
+
+- Write/update path: `src/sse/services/auth.ts::markAccountUnavailable()`
+- Account selection/filtering: `src/sse/services/auth.ts::getProviderCredentials...`
+- Cooldown calculation: `open-sse/services/accountFallback.ts::checkFallbackError()`
+- Settings: `src/lib/resilience/settings.ts`
+
+Important fields on provider connections:
+
+```ts
+rateLimitedUntil;
+testStatus: "unavailable";
+lastError;
+lastErrorType;
+errorCode;
+backoffLevel;
+```
+
+During account selection, a connection is skipped while:
+
+```ts
+new Date(rateLimitedUntil).getTime() > Date.now();
+```
+
+Cooldowns are also lazy: when `rateLimitedUntil` is in the past, the connection becomes
+eligible again. On successful use, `clearAccountError()` clears `testStatus`,
+`rateLimitedUntil`, error fields, and `backoffLevel`.
+
+Default connection cooldown behavior:
+
+- OAuth base cooldown: `5s`.
+- API-key base cooldown: `3s`.
+- API-key `429` should prefer upstream retry hints (`Retry-After`, reset headers, or
+  parseable reset text) when available.
+- Repeated recoverable failures use exponential backoff:
+
+```ts
+baseCooldownMs * 2 ** failureIndex;
+```
+
+The anti-thundering-herd guard prevents concurrent failures on the same connection from
+repeatedly extending the cooldown or double-incrementing `backoffLevel`.
+
+Terminal states are not cooldowns. `banned`, `expired`, and `credits_exhausted` are
+intended to stay unavailable until credentials/settings change or an operator resets
+them. Do not overwrite terminal states with transient cooldown state.
+
+### Model Lockout
+
+**Scope**: provider + connection + model.
+
+**Purpose**: avoid disabling a whole connection when only one model is unavailable or
+quota-limited for that connection.
+
+Examples:
+
+- Per-model quota providers returning `429`.
+- Local providers returning `404` for one missing model.
+- Provider-specific mode/model permission failures such as selected Grok modes.
+
+Model lockout lives in `open-sse/services/accountFallback.ts` and lets the same
+connection continue serving other models.
+
+### Debugging Guidance
+
+- If all keys for a provider are skipped, inspect both provider breaker state and each
+  connection's `rateLimitedUntil`/`testStatus`.
+- If a provider appears permanently excluded after the reset window, check whether code
+  is reading raw `state` instead of using `getStatus()`/`canExecute()`.
+- If one provider key fails but others should work, prefer connection cooldown over
+  provider breaker.
+- If only one model fails, prefer model lockout over connection cooldown.
+- If a state should self-recover, it should have a future timestamp/reset timeout and a
+  read path that refreshes expired state. Permanent statuses require manual credential
+  or config changes.
+
+---
+
+## Key Conventions
+
+### Code Style
+
+- **2 spaces**, semicolons, double quotes, 100 char width, es5 trailing commas (enforced by lint-staged via Prettier)
+- **Imports**: external → internal (`@/`, `@omniroute/open-sse`) → relative
+- **Naming**: files=camelCase/kebab, components=PascalCase, constants=UPPER_SNAKE
+- **ESLint**: `no-eval`, `no-implied-eval`, `no-new-func` = error everywhere; `no-explicit-any` = warn in `open-sse/` and `tests/`
+- **TypeScript**: `strict: false`, target ES2022, module esnext, resolution bundler. Prefer explicit types.
+
+### Database
+
+- **Always** go through `src/lib/db/` domain modules — **never** write raw SQL in routes or handlers
+- **Never** add logic to `src/lib/localDb.ts` (re-export layer only)
+- **Never** barrel-import from `localDb.ts` — import specific `db/` modules instead
+- DB singleton: `getDbInstance()` from `src/lib/db/core.ts` (WAL journaling)
+- Migrations: `src/lib/db/migrations/` — versioned SQL files, idempotent, run in transactions
+
+### Error Handling
+
+- try/catch with specific error types, log with pino context
+- Never swallow errors in SSE streams — use abort signals for cleanup
+- Return proper HTTP status codes (4xx/5xx)
+
+### Security
+
+- **Never** use `eval()`, `new Function()`, or implied eval
+- Validate all inputs with Zod schemas
+- Encrypt credentials at rest (AES-256-GCM)
+- Upstream header denylist: `src/shared/constants/upstreamHeaders.ts` — keep sanitize, Zod schemas, and unit tests aligned when editing
+- **Public upstream credentials** (Gemini/Antigravity/Windsurf-style OAuth client_id/secret + Firebase Web keys extracted from public CLIs): **MUST** be embedded via `resolvePublicCred()` from `open-sse/utils/publicCreds.ts` — **never** as string literals. See `docs/security/PUBLIC_CREDS.md` for the mandatory pattern.
+- **Error responses** (HTTP / SSE / executor / MCP handler): **MUST** route through `buildErrorBody()` or `sanitizeErrorMessage()` from `open-sse/utils/error.ts` — **never** put raw `err.stack` or `err.message` in a response body. See `docs/security/ERROR_SANITIZATION.md`.
+- **Shell commands built from variables**: when calling `exec()`/`spawn()` with a script that needs runtime values, pass them via the `env` option (shell-escaped automatically) — **never** string-interpolate untrusted/external paths into the script body. Reference: `src/mitm/cert/install.ts::updateNssDatabases`.
+- **Secure-by-default libraries** ([tldrsec/awesome-secure-defaults](https://github.com/tldrsec/awesome-secure-defaults)): prefer Helmet.js, DOMPurify, ssrf-req-filter, safe-regex, Google Tink over custom implementations whenever adding new security-sensitive surfaces.
+
+---
+
+## Common Modification Scenarios
+
+### Adding a New Provider
+
+1. Register in `src/shared/constants/providers.ts` (Zod-validated at load)
+2. Add executor in `open-sse/executors/` if custom logic needed (extend `BaseExecutor`)
+3. Add translator in `open-sse/translator/` if non-OpenAI format
+4. Add OAuth config in `src/lib/oauth/constants/oauth.ts` if OAuth-based — if the upstream CLI ships a public client_id/secret, embed via `resolvePublicCred()` (see `docs/security/PUBLIC_CREDS.md`), **never** as a literal
+5. Register models in `open-sse/config/providerRegistry.ts`
+6. Write tests in `tests/unit/` (include the publicCreds shape assertion if you added a new embedded default)
+
+### Adding a New API Route
+
+1. Create directory under `src/app/api/v1/your-route/`
+2. Create `route.ts` with `GET`/`POST` handlers
+3. Follow pattern: CORS → Zod body validation → optional auth → handler delegation
+4. Handler goes in `open-sse/handlers/` (import from there, not inline)
+5. Error responses use `buildErrorBody()` / `errorResponse()` from `open-sse/utils/error.ts` (auto-sanitized — never put `err.stack` or `err.message` raw in the body). See `docs/security/ERROR_SANITIZATION.md`.
+6. Add tests — including at least one assertion that error responses do not leak stack traces (`!body.error.message.includes("at /")`)
+
+### Adding a New DB Module
+
+1. Create `src/lib/db/yourModule.ts` — import `getDbInstance` from `./core.ts`
+2. Export CRUD functions for your domain table(s)
+3. Add migration in `src/lib/db/migrations/` if new tables needed
+4. Re-export from `src/lib/localDb.ts` (add to the re-export list only)
+5. Write tests
+
+### Adding a New MCP Tool
+
+1. Add tool definition in `open-sse/mcp-server/tools/` with Zod input schema + async handler
+2. Register in tool set (wired by `createMcpServer()`)
+3. Assign to appropriate scope(s)
+4. Write tests (tool invocation logged to `mcp_audit` table)
+
+### Adding a New A2A Skill
+
+1. Create skill in `src/lib/a2a/skills/` (5 already exist: smart-routing, quota-management, provider-discovery, cost-analysis, health-report)
+2. Skill receives task context (messages, metadata) → returns structured result
+3. Register in `A2A_SKILL_HANDLERS` in `src/lib/a2a/taskExecution.ts`
+4. Expose in `src/app/.well-known/agent.json/route.ts` (Agent Card)
+5. Write tests in `tests/unit/`
+6. Document in `docs/frameworks/A2A-SERVER.md` skill table
+
+### Adding a New Cloud Agent
+
+1. Create agent class in `src/lib/cloudAgent/agents/` extending `CloudAgentBase` (3 already exist: codex-cloud, devin, jules)
+2. Implement `createTask`, `getStatus`, `approvePlan`, `sendMessage`, `listSources`
+3. Register in `src/lib/cloudAgent/registry.ts`
+4. Add OAuth/credentials handling if needed (`src/lib/oauth/providers/`)
+5. Tests + document in `docs/frameworks/CLOUD_AGENT.md`
+
+### Adding a New Embedded Service
+
+1. Create installer in `src/lib/services/installers/{name}.ts` modeled on `ninerouter.ts` (use `runNpm` from `installers/utils.ts` — no shell interpolation, hard rule #13).
+2. Register the service in `src/lib/services/bootstrap.ts` (add to `SERVICES[]` array and extend `buildSpawnArgsFactory()`).
+3. Add a DB seed row for the new service in `src/lib/db/migrations/` (`version_manager` table, `status='not_installed'`, `auto_start=0`).
+4. Create 7 API endpoints under `src/app/api/services/{name}/` (`_lib.ts`, `install`, `start`, `stop`, `restart`, `update`, `status`, `auto-start`). All delegate errors through `createErrorResponse()`. The shared `logs` endpoint is already wired via `[name]/logs/route.ts`.
+5. Verify `/api/services/` is in `LOCAL_ONLY_API_PREFIXES` in `src/server/authz/routeGuard.ts`; add a test asserting `isLocalOnlyPath()` returns `true` for the new prefix if you add one (hard rule #17).
+6. Add a UI tab in `src/app/(dashboard)/dashboard/providers/services/tabs/` reusing `ServiceStatusCard`, `ServiceLifecycleButtons`, `ServiceLogsPanel`.
+7. Document in `docs/frameworks/EMBEDDED-SERVICES.md` (update §1 service table + §4 API reference) and `docs/openapi.yaml`.
+8. Write tests: unit (`tests/unit/services/`), integration (`tests/integration/services/`, gated by `RUN_SERVICES_INT=1`), and update `docs/ops/RELEASE_CHECKLIST.md` smoke section.
+
+### Adding a New Guardrail / Eval / Skill / Webhook event
+
+- Guardrail: `src/lib/guardrails/` → docs: `docs/security/GUARDRAILS.md`
+- Eval suite: `src/lib/evals/` → docs: `docs/frameworks/EVALS.md`
+- Skill (sandbox): `src/lib/skills/` → docs: `docs/frameworks/SKILLS.md`
+- Webhook event: `src/lib/webhookDispatcher.ts` → docs: `docs/frameworks/WEBHOOKS.md`
+
+---
+
+## Reference Documentation
+
+For any non-trivial change, read the matching deep-dive first:
+
+| Area                                          | Doc                                                               |
+| --------------------------------------------- | ----------------------------------------------------------------- |
+| Repo navigation                               | `docs/architecture/REPOSITORY_MAP.md`                             |
+| Architecture                                  | `docs/architecture/ARCHITECTURE.md`                               |
+| Engineering reference                         | `docs/architecture/CODEBASE_DOCUMENTATION.md`                     |
+| Auto-Combo (9-factor scoring, 15 strategies)  | `docs/routing/AUTO-COMBO.md`                                      |
+| Resilience (3 mechanisms)                     | `docs/architecture/RESILIENCE_GUIDE.md`                           |
+| Reasoning replay                              | `docs/routing/REASONING_REPLAY.md`                                |
+| Skills framework                              | `docs/frameworks/SKILLS.md`                                       |
+| Memory system (FTS5 + Qdrant)                 | `docs/frameworks/MEMORY.md`                                       |
+| Cloud agents                                  | `docs/frameworks/CLOUD_AGENT.md`                                  |
+| Guardrails (PII / injection / vision)         | `docs/security/GUARDRAILS.md`                                     |
+| Public upstream credentials (Gemini/etc.)     | `docs/security/PUBLIC_CREDS.md`                                   |
+| Error message sanitization                    | `docs/security/ERROR_SANITIZATION.md`                             |
+| Evals                                         | `docs/frameworks/EVALS.md`                                        |
+| Compliance / audit                            | `docs/security/COMPLIANCE.md`                                     |
+| Webhooks                                      | `docs/frameworks/WEBHOOKS.md`                                     |
+| Authorization pipeline                        | `docs/architecture/AUTHZ_GUIDE.md`                                |
+| Stealth (TLS / fingerprint)                   | `docs/security/STEALTH_GUIDE.md`                                  |
+| Agent protocols (A2A / ACP / Cloud)           | `docs/frameworks/AGENT_PROTOCOLS_GUIDE.md`                        |
+| MCP server                                    | `docs/frameworks/MCP-SERVER.md`                                   |
+| A2A server                                    | `docs/frameworks/A2A-SERVER.md`                                   |
+| API reference + OpenAPI                       | `docs/reference/API_REFERENCE.md` + `docs/openapi.yaml` |
+| Provider catalog (auto-generated)             | `docs/reference/PROVIDER_REFERENCE.md`                            |
+| Release flow                                  | `docs/ops/RELEASE_CHECKLIST.md`                                   |
+| Embedded services                             | `docs/frameworks/EMBEDDED-SERVICES.md`                            |
+| Quality gates (~48 scripts, allowlist policy) | `docs/architecture/QUALITY_GATES.md`                              |
+
+---
+
+## Testing
+
+| What                    | Command                                                                     |
+| ----------------------- | --------------------------------------------------------------------------- |
+| Unit tests              | `npm run test:unit`                                                         |
+| Single file             | `node --import tsx/esm --test tests/unit/file.test.ts`                      |
+| Vitest (MCP, autoCombo) | `npm run test:vitest`                                                       |
+| E2E (Playwright)        | `npm run test:e2e`                                                          |
+| Protocol E2E (MCP+A2A)  | `npm run test:protocols:e2e`                                                |
+| Ecosystem               | `npm run test:ecosystem`                                                    |
+| Coverage gate           | `npm run test:coverage` (60/60/60/60 — statements/lines/functions/branches) |
+| Coverage report         | `npm run coverage:report`                                                   |
+
+**PR rule**: If you change production code in `src/`, `open-sse/`, `electron/`, or `bin/`, you must include or update tests in the same PR.
+
+**Test layer preference**: unit first → integration (multi-module or DB state) → e2e (UI/workflow only). Encode bug reproductions as automated tests before or alongside the fix.
+
+**Both test runners must pass**: `npm run test:unit` (Node native — most tests) AND `npm run test:vitest` (MCP server, autoCombo, cache) cover **non-overlapping files**. Both are wired in CI (jobs `test-unit` and `test-vitest`) and must be green before merging. A PR where only one suite passes may silently ship broken MCP tools or routing regressions.
+
+**Bug fix / issue triage protocol (Hard Rule #18)**: Every fix for a reported issue must be validated by one of the following — no exceptions:
+
+1. **TDD (preferred)** — write a failing test reproducing the bug → fix it → confirm the test passes. The test becomes the permanent regression guard. Touch only the files the test proves need changing; nothing more.
+2. **Real-environment test (when TDD is not possible)** — deploy to the production VPS (`root@192.168.0.15`) and run a documented live test. Record the exact command + result in the PR description. Applies to: OAuth upstream flows, Cloudflare/WS upstream behavior, UI-only regressions, hardware-dependent behavior.
+3. "It worked locally without a test" does not count. A fix without a test or a VPS validation record is not a fix — it is a guess.
+
+Why this matters: fixing bug A while opening bug B is worse than not fixing at all. The TDD/VPS gate enforces surgical scope — you touch only what the failing test proves is broken. Examples where this paid off: #3090 (claude-web 403), #3113 (WS HTTP fallback), #3052 (heap-guard auto-calibration).
+
+**Copilot coverage policy**: When a PR changes production code and coverage is below 60% (statements/lines/functions/branches), do not just report — add or update tests, rerun the coverage gate, then ask for confirmation. Include commands run, changed test files, and final coverage result in the PR report.
+
+---
+
+## Git Workflow
+
+```bash
+# Never commit directly to main
+git checkout -b feat/your-feature
+git commit -m "feat: describe your change"
+git push -u origin feat/your-feature
+```
+
+**Branch prefixes**: `feat/`, `fix/`, `refactor/`, `docs/`, `test/`, `chore/`
+
+**Commit format** (Conventional Commits): `feat(db): add circuit breaker` — scopes: `db`, `sse`, `oauth`, `dashboard`, `api`, `cli`, `docker`, `ci`, `mcp`, `a2a`, `memory`, `skills`
+
+**Husky hooks**:
+
+- **pre-commit**: lint-staged + `check-docs-sync` + `check:any-budget:t11`
+- **pre-push**: fast deterministic gates (`check:any-budget:t11` + `check:tracked-artifacts`); intentionally excludes `test:unit` (slow — covered by the CI `test-unit` job). Activated 2026-06-13 (Quality Gates Fase 6A.12).
+
+### Worktree isolation (MANDATORY for every development task)
+
+Multiple sessions/agents work this repo in parallel. The main checkout is **shared**, so a
+`git checkout`/branch switch in it silently discards another session's uncommitted work and
+yanks the branch out from under whatever else is running (incidents: 2026-06-05, 2026-06-13).
+
+**Rule: never develop on the shared main checkout. Every task gets its own git worktree on its
+own dedicated branch, and you MUST confirm the base branch with the operator before creating it.**
+
+1. **Ask first — which base branch?** Before creating anything, ask the operator (via
+   `AskUserQuestion`, unless they already told you) from which branch the new worktree/branch
+   should be cut. Do NOT assume `main` or "whatever I'm on" — the answer is usually the active
+   `release/vX.Y.Z`, but it can be another feature/release branch. Get the base explicitly.
+2. **Create an isolated worktree + branch off that base** (never reuse the main checkout):
+
+   ```bash
+   BASE_BRANCH="release/vX.Y.Z"          # ← the branch the operator confirmed in step 1
+   TASK="feat/your-feature"               # feat/ fix/ refactor/ docs/ test/ chore/
+   git fetch origin "$BASE_BRANCH"
+   git worktree add ".worktrees/${TASK##*/}" -b "$TASK" "origin/$BASE_BRANCH"
+   cd ".worktrees/${TASK##*/}"
+   # symlink node_modules from the main checkout to skip a per-worktree npm install:
+   ln -s "$(git -C <main_checkout> rev-parse --show-toplevel)/node_modules" node_modules
+   ```
+
+   In Claude Code prefer the native `EnterWorktree` tool (create the worktree with the command
+   above, then call `EnterWorktree` with its `path`).
+
+3. **Work, commit, push, open the PR — all from inside the worktree.** Never `git checkout` a
+   different branch inside a worktree another session might share.
+4. **Tear down only your own** worktree + branch when done, from the main checkout:
+   `git worktree remove .worktrees/<dir>` then `git branch -D <task>`. Never blanket-delete
+   `fix/*`/`feat/*` — other sessions keep their own; delete only the branches you created, by name.
+5. **Never touch another session's worktree, branch, or uncommitted changes.** If `git worktree
+list` shows worktrees you didn't create, leave them alone. End every session with the main
+   checkout back on the branch it started on (the active `release/vX.Y.Z`, never `main`).
+
+---
+
+## Environment
+
+- **Runtime**: Node.js ≥22.0.0 <23 || ≥24.0.0 <27, ES Modules
+- **TypeScript**: 6.0+, target ES2022, module esnext, resolution bundler
+- **Path aliases**: `@/*` → `src/`, `@omniroute/open-sse` → `open-sse/`, `@omniroute/open-sse/*` → `open-sse/*`
+- **Default port**: 20128 (API + dashboard on same port)
+- **Data directory**: `DATA_DIR` env var, defaults to `~/.omniroute/`
+- **Key env vars**: `PORT`, `JWT_SECRET`, `API_KEY_SECRET`, `INITIAL_PASSWORD`, `REQUIRE_API_KEY`, `APP_LOG_LEVEL`
+- Setup: `cp .env.example .env` then generate `JWT_SECRET` (`openssl rand -base64 48`) and `API_KEY_SECRET` (`openssl rand -hex 32`)
+
+---
+
+## Quality Gates & Ratchets
+
+OmniRoute has **~48 quality-gate scripts** (`scripts/check/` + `scripts/quality/`) wired
+across **9 gate-running jobs** in `.github/workflows/ci.yml` (`lint`, `quality-gate`,
+`quality-extended`, `docs-sync-strict`, `i18n-ui-coverage`, `i18n`, `pr-test-policy`,
+`test-vitest`, `sonarqube`), plus the `quality.yml` fast-gates job (PR→`release/**`) and
+3 nightly workflows (`nightly-property`, `nightly-resilience`, `nightly-llm-security`;
+`nightly-mutation` once merged). Full inventory, per-job breakdown, and operational
+procedures are in [`docs/architecture/QUALITY_GATES.md`](docs/architecture/QUALITY_GATES.md).
+
+**Quick reference:**
+
+- Gates in jobs `lint` + `docs-sync-strict`: pass/fail policy gates —
+  fix the violation or add an allowlist entry with a justification comment + tracking issue.
+- Gates in job `quality-gate`: ratchet — metrics (ESLint warnings, code coverage, duplication,
+  complexity) must not regress vs `quality-baseline.json`. Update via
+  `npm run quality:ratchet -- --update` when a metric genuinely improves.
+- Job `test-vitest` runs `npm run test:vitest` (MCP tools, autoCombo, cache) — blocking.
+  `test:vitest:ui` is advisory until UI component tests are triaged.
+
+**Allowlist policy (short form):** Fix the cause; use the allowlist only for pre-existing
+violations you cannot fix in the same PR. Add a comment with justification + issue number.
+Stale allowlist entries (suppressing a violation that no longer exists) will be caught by
+the stale-enforcement added in Fase 6A.3.
+
+---
+
+## Hard Rules
+
+1. Never commit secrets or credentials
+2. Never add logic to `localDb.ts`
+3. Never use `eval()` / `new Function()` / implied eval
+4. Never commit directly to `main`
+5. Never write raw SQL in routes — use `src/lib/db/` modules
+6. Never silently swallow errors in SSE streams
+7. Always validate inputs with Zod schemas
+8. Always include tests when changing production code
+9. Coverage must not regress below the baseline frozen in `quality-baseline.json` (ratchet); absolute floor is 60% (statements/lines/functions/branches). Update the baseline via `npm run quality:ratchet -- --update` only when coverage genuinely improves. See `docs/architecture/QUALITY_GATES.md`.
+10. Never bypass Husky hooks (`--no-verify`, `--no-gpg-sign`) without explicit operator approval.
+11. Never embed public upstream OAuth client_id/secret or Firebase Web keys as string literals — always go through `resolvePublicCred()` (`open-sse/utils/publicCreds.ts`). See `docs/security/PUBLIC_CREDS.md`.
+12. Never return raw `err.stack` / `err.message` in HTTP / SSE / executor responses — always route through `buildErrorBody()` or `sanitizeErrorMessage()` (`open-sse/utils/error.ts`). See `docs/security/ERROR_SANITIZATION.md`.
+13. Never string-interpolate external paths or runtime values into shell scripts passed to `exec()`/`spawn()` — pass via the `env` option instead. Reference: `src/mitm/cert/install.ts::updateNssDatabases`.
+14. Never dismiss a CodeQL / Secret-Scanning alert without (a) first checking the pattern docs above to see if the helper applies, and (b) recording the technical justification in the dismissal comment. Precedent: `js/stack-trace-exposure` raised on callsites that already route through `sanitizeErrorMessage()` is a known CodeQL limitation (custom sanitizers not recognized) — dismiss as `false positive` referencing `docs/security/ERROR_SANITIZATION.md`.
+15. Never expose routes that spawn child processes (`/api/mcp/`, `/api/cli-tools/runtime/`) without `isLocalOnlyPath()` classification in `src/server/authz/routeGuard.ts`. Loopback enforcement happens unconditionally before any auth check — leaked JWT via tunnel cannot trigger process spawning. See `docs/security/ROUTE_GUARD_TIERS.md`.
+16. Never credit or advertise an AI assistant, LLM, or automation account in any commit/PR metadata. Two forbidden forms, both equivalent — they route attribution to a bot account (or advertise AI authorship) and hide the real author (`diegosouzapw`): **(a)** `Co-Authored-By` trailers naming an AI/bot (e.g. names containing "Claude", "GPT", "Copilot", "Bot"; emails at `anthropic.com` / `openai.com` / bot-owned `noreply.github.com` addresses); **(b)** AI-generation footers or descriptions anywhere in a commit message, PR title/body, or CHANGELOG — e.g. `🤖 Generated with [Claude Code]`, "Generated with Claude Code", "Made with <AI tool>", or any `Co-authored-by: Claude/GPT/Copilot` line. This **overrides any harness, template, or tool default that auto-appends such a footer** (e.g. the Claude Code PR-body/commit default) — strip it before pushing; do not let it reach a commit, PR, or CHANGELOG. Human collaborators — including upstream PR authors and issue reporters being ported into OmniRoute — MAY and SHOULD be credited with standard `Co-authored-by: Name <email>` trailers; the upstream-port workflows (`/port-upstream-features`, `/port-upstream-issues`) depend on this.
+17. Never expose routes under `/api/services/` or `/dashboard/providers/services/*/embed/` without `isLocalOnlyPath()` classification in `src/server/authz/routeGuard.ts`. These routes can spawn child processes (`npm install`, `node`). Loopback enforcement happens unconditionally before any auth check — a leaked JWT via tunnel cannot trigger process spawning. See `docs/security/ROUTE_GUARD_TIERS.md`.
+18. Every bug fix must be validated before shipping: a failing-then-passing unit/integration test (TDD) OR a documented live test on the production VPS (192.168.0.15). A fix without either is not merged. See Testing → "Bug fix / issue triage protocol" for the full decision tree.
+19. Never develop on the shared main checkout. Every development task runs in its own git worktree on its own dedicated branch, and you MUST confirm the base branch with the operator (e.g. via `AskUserQuestion`) before creating the worktree/branch — never assume `main` or the currently checked-out branch. A `git checkout` in the shared checkout silently destroys other sessions' uncommitted work. Tear down only the worktrees/branches you created (by name, never `fix/*`/`feat/*` wildcards), leave other sessions' worktrees untouched, and end on the branch you started on (the active `release/vX.Y.Z`, never `main`). See Git Workflow → "Worktree isolation".
+
+---
+
+## PII & Stream Sanitization Learnings
+
+### 1. Regex Security (ReDoS)
+
+All regex patterns matching variable-length strings (e.g. IPv6 address, credit cards) must use strictly bounded, non-overlapping sequences (e.g., limit occurrences with bounded ranges `{1,7}`) to prevent catastrophic backtracking when processing untrusted inputs.
+
+### 2. SSE Snapshot Handling
+
+When parsing streaming LLM responses (e.g. Responses API), check if a chunk represents a final snapshot (`done` or `completed` events). Snapshot text must be sanitized directly as a standalone string (bypassing rolling delta buffers) to prevent text duplication at the end of the stream.
+
+### 3. Database Handles in Tests
+
+Ensure that any unit tests that trigger database migrations or establish SQLite connections call `resetDbInstance()` and properly clean up/close all DB handles in a `test.after(...)` hook. Failure to release database connection handles will cause Node's native test runner to hang indefinitely.

CODE_OF_CONDUCT.md

@@ -0,0 +1,131 @@
+# Contributor Covenant Code of Conduct
+
+## Our Pledge
+
+We as members, contributors, and leaders pledge to make participation in our
+community a harassment-free experience for everyone, regardless of age, body
+size, visible or invisible disability, ethnicity, sex characteristics, gender
+identity and expression, level of experience, education, socio-economic status,
+nationality, personal appearance, race, religion, or sexual identity
+and orientation.
+
+We pledge to act and interact in ways that contribute to an open, welcoming,
+diverse, inclusive, and healthy community.
+
+## Our Standards
+
+Examples of behavior that contributes to a positive environment for our
+community include:
+
+- Demonstrating empathy and kindness toward other people
+- Being respectful of differing opinions, viewpoints, and experiences
+- Giving and gracefully accepting constructive feedback
+- Accepting responsibility and apologizing to those affected by our mistakes,
+  and learning from the experience
+- Focusing on what is best not just for us as individuals, but for the
+  overall community
+
+Examples of unacceptable behavior include:
+
+- The use of sexualized language or imagery, and sexual attention or
+  advances of any kind
+- Trolling, insulting or derogatory comments, and personal or political attacks
+- Public or private harassment
+- Publishing others' private information, such as a physical or email
+  address, without their explicit permission
+- Other conduct which could reasonably be considered inappropriate in a
+  professional setting
+
+## Enforcement Responsibilities
+
+Community leaders are responsible for clarifying and enforcing our standards of
+acceptable behavior and will take appropriate and fair corrective action in
+response to any behavior that they deem inappropriate, threatening, offensive,
+or harmful.
+
+Community leaders have the right and responsibility to remove, edit, or reject
+comments, commits, code, wiki edits, issues, and other contributions that are
+not aligned to this Code of Conduct, and will communicate reasons for moderation
+decisions when appropriate.
+
+## Scope
+
+This Code of Conduct applies within all community spaces, and also applies when
+an individual is officially representing the community in public spaces.
+Examples of representing our community include using an official e-mail address,
+posting via an official social media account, or acting as an appointed
+representative at an online or offline event.
+
+## Enforcement
+
+Instances of abusive, harassing, or otherwise unacceptable behavior may be
+reported to the community leaders responsible for enforcement by opening a
+private security advisory at
+<https://github.com/diegosouzapw/OmniRoute/security/advisories/new>
+or by emailing the maintainer at diegosouza.pw@outlook.com.
+For security-sensitive incidents, see [`SECURITY.md`](SECURITY.md).
+All complaints will be reviewed and investigated promptly and fairly.
+
+All community leaders are obligated to respect the privacy and security of the
+reporter of any incident.
+
+## Enforcement Guidelines
+
+Community leaders will follow these Community Impact Guidelines in determining
+the consequences for any action they deem in violation of this Code of Conduct:
+
+### 1. Correction
+
+**Community Impact**: Use of inappropriate language or other behavior deemed
+unprofessional or unwelcome in the community.
+
+**Consequence**: A private, written warning from community leaders, providing
+clarity around the nature of the violation and an explanation of why the
+behavior was inappropriate. A public apology may be requested.
+
+### 2. Warning
+
+**Community Impact**: A violation through a single incident or series
+of actions.
+
+**Consequence**: A warning with consequences for continued behavior. No
+interaction with the people involved, including unsolicited interaction with
+those enforcing the Code of Conduct, for a specified period of time. This
+includes avoiding interactions in community spaces as well as external channels
+like social media. Violating these terms may lead to a temporary or
+permanent ban.
+
+### 3. Temporary Ban
+
+**Community Impact**: A serious violation of community standards, including
+sustained inappropriate behavior.
+
+**Consequence**: A temporary ban from any sort of interaction or public
+communication with the community for a specified period of time. No public or
+private interaction with the people involved, including unsolicited interaction
+with those enforcing the Code of Conduct, is allowed during this period.
+Violating these terms may lead to a permanent ban.
+
+### 4. Permanent Ban
+
+**Community Impact**: Demonstrating a pattern of violation of community
+standards, including sustained inappropriate behavior, harassment of an
+individual, or aggression toward or disparagement of classes of individuals.
+
+**Consequence**: A permanent ban from any sort of public interaction within
+the community.
+
+## Attribution
+
+This Code of Conduct is adapted from the [Contributor Covenant][homepage],
+version 2.1, available at
+https://www.contributor-covenant.org/version/2/1/code_of_conduct.html.
+
+Community Impact Guidelines were inspired by [Mozilla's code of conduct
+enforcement ladder](https://github.com/mozilla/diversity).
+
+[homepage]: https://www.contributor-covenant.org
+
+For answers to common questions about this code of conduct, see the FAQ at
+https://www.contributor-covenant.org/faq. Translations are available at
+https://www.contributor-covenant.org/translations.

CONTRIBUTING.md

@@ -8,7 +8,7 @@ Thank you for your interest in contributing! This guide covers everything you ne
 
 ### Prerequisites
 
-- **Node.js** 20+ (recommended: 22 LTS)
+- **Node.js** `>=22.22.3 <23`, or `>=24.0.0 <27` (recommended: 24 LTS)
 - **npm** 10+
 - **Git**
 
@@ -33,13 +33,24 @@ echo "API_KEY_SECRET=$(openssl rand -hex 32)" >> .env
 
 Key variables for development:
 
-| Variable               | Development Default     | Description               |
-| ---------------------- | ----------------------- | ------------------------- |
-| `PORT`                 | `3000`                  | Server port               |
-| `NEXT_PUBLIC_BASE_URL` | `http://localhost:3000` | Base URL for frontend     |
-| `JWT_SECRET`           | (generate above)        | JWT signing secret        |
-| `INITIAL_PASSWORD`     | `123456`                | First login password      |
-| `ENABLE_REQUEST_LOGS`  | `false`                 | Enable debug request logs |
+| Variable               | Development Default      | Description           |
+| ---------------------- | ------------------------ | --------------------- |
+| `PORT`                 | `20128`                  | Server port           |
+| `NEXT_PUBLIC_BASE_URL` | `http://localhost:20128` | Base URL for frontend |
+| `JWT_SECRET`           | (generate above)         | JWT signing secret    |
+| `INITIAL_PASSWORD`     | `CHANGEME`               | First login password  |
+| `APP_LOG_LEVEL`        | `info`                   | Log verbosity level   |
+
+### Dashboard Settings
+
+The dashboard provides UI toggles for features that can also be configured via environment variables:
+
+| Setting Location    | Toggle             | Description                    |
+| ------------------- | ------------------ | ------------------------------ |
+| Settings → Advanced | Debug Mode         | Enable debug request logs (UI) |
+| Settings → General  | Sidebar Visibility | Show/hide sidebar sections     |
+
+These settings are stored in the database and persist across restarts, overriding env var defaults when set.
 
 ### Running Locally
 
@@ -48,17 +59,44 @@ Key variables for development:
 npm run dev
 
 # Production build
-npm run build
+npm run build    # next build → .build/next/ then assembleStandalone → dist/
 npm run start
 
+# Release build (clean rebuild + HEAD sentinel — required for deploy)
+npm run build:release   # rm -rf .build dist && build + writes dist/BUILD_SHA
+
 # Common port configuration
 PORT=20128 NEXT_PUBLIC_BASE_URL=http://localhost:20128 npm run dev
 ```
 
+### Build Output Layout
+
+| Directory | Contents                                                                  | Tracked |
+| --------- | ------------------------------------------------------------------------- | ------- |
+| `src/`    | Application source (TypeScript / TSX)                                     | Yes     |
+| `.build/` | Intermediates — `next build` output (gitignored, `distDir = .build/next`) | No      |
+| `dist/`   | Shippable bundle — assembled by `assembleStandalone` (gitignored)         | No      |
+
+The build pipeline is a single pass:
+
+```
+npm run build
+  └─ next build → .build/next/standalone  (Next.js output)
+  └─ assembleStandalone()                 (copies standalone + static + public + native assets)
+       └─ output: dist/                   (server.js, .next/static/, public/, node_modules/)
+```
+
+`npm run build:release` additionally cleans both directories first and writes
+`dist/BUILD_SHA` (= `git rev-parse --short HEAD`) as a deploy integrity sentinel.
+
+> **VPS deploy note:** the remote image directory `/usr/lib/node_modules/omniroute/app/`
+> is unchanged. The deploy skills rsync the contents of `dist/` into it.
+> Only the in-repo build output path moved (`app/` → `dist/`).
+
 Default URLs:
 
-- **Dashboard**: `http://localhost:3000/dashboard`
-- **API**: `http://localhost:3000/v1`
+- **Dashboard**: `http://localhost:20128/dashboard`
+- **API**: `http://localhost:20128/v1`
 
 ---
 
@@ -97,50 +135,80 @@ test: add observability unit tests
 refactor(db): consolidate rate limit tables
 ```
 
-Scopes: `db`, `sse`, `oauth`, `dashboard`, `api`, `cli`, `docker`, `ci`.
+Scopes (v3.8): `db`, `sse`, `oauth`, `dashboard`, `api`, `cli`, `docker`, `ci`, `mcp`, `a2a`, `memory`, `skills`, `cloud-agent`, `guardrails`, `compression`, `auto-combo`, `resilience`, `providers`, `executors`, `translator`, `domain`, `authz`.
 
 ---
 
 ## Running Tests
 
 ```bash
-# All unit tests
-npm test
-npm run test:unit
+# All tests (unit + vitest + ecosystem + e2e)
+npm run test:all
 
-# Specific test suites
-npm run test:security     # Security tests
-npm run test:fixes        # Fix verification tests
+# Single test file (Node.js native test runner — most tests use this)
+node --import tsx/esm --test tests/unit/your-file.test.ts
 
-# With coverage
-npm run test:coverage
+# Vitest (MCP server, autoCombo, cache)
+npm run test:vitest
 
 # E2E tests (requires Playwright)
 npm run test:e2e
 
+# Protocol clients E2E (MCP transports, A2A)
+npm run test:protocols:e2e
+
+# Ecosystem compatibility tests
+npm run test:ecosystem
+
+# Coverage gate: 75% statements/lines/functions, 70% branches
+npm run test:coverage
+npm run coverage:report
+
 # Lint + format check
 npm run lint
 npm run check
 ```
 
-Current test status: **368+ unit tests** covering:
+Coverage notes:
+
+- `npm run test:coverage` measures source coverage for the main unit test suite, excludes `tests/**`, and includes `open-sse/**`
+- Pull requests must keep the coverage gate at **75%+** statements/lines/functions and **70%+** branches
+- If a PR changes production code in `src/`, `open-sse/`, `electron/`, or `bin/`, it must add or update automated tests in the same PR
+- `npm run coverage:report` prints the detailed file-by-file report from the latest coverage run
+- `npm run test:coverage:legacy` preserves the older metric for historical comparison
+- See `docs/ops/COVERAGE_PLAN.md` for the phased coverage improvement roadmap
+
+### Pull Request Requirements
+
+Before opening or merging a PR:
+
+- Run `npm run test:unit`
+- Run `npm run test:coverage`
+- Ensure the coverage gate stays at **75%+** statements/lines/functions, **70%+** branches
+- Include the changed or added test files in the PR description when production code changed
+- Check the SonarQube result on the PR when the project secrets are configured in CI
+
+Current test status: **122 unit test files** covering:
 
 - Provider translators and format conversion
 - Rate limiting, circuit breaker, and resilience
 - Semantic cache, idempotency, progress tracking
-- Database operations and schema
+- Database operations and schema (21 DB modules)
 - OAuth flows and authentication
-- API endpoint validation
+- API endpoint validation (Zod v4)
+- MCP server tools and scope enforcement
+- Memory and Skills systems
 
 ---
 
 ## Code Style
 
 - **ESLint** — Run `npm run lint` before committing
-- **Prettier** — Auto-formatted via `lint-staged` on commit
-- **TypeScript** — All `src/` code uses `.ts`/`.tsx`; document with TSDoc (`@param`, `@returns`, `@throws`)
+- **Prettier** — Auto-formatted via `lint-staged` on commit (2 spaces, semicolons, double quotes, 100 char width, es5 trailing commas)
+- **TypeScript** — All `src/` code uses `.ts`/`.tsx`; `open-sse/` uses `.ts`/`.js`; document with TSDoc (`@param`, `@returns`, `@throws`)
 - **No `eval()`** — ESLint enforces `no-eval`, `no-implied-eval`, `no-new-func`
-- **Zod validation** — Use Zod schemas for API input validation
+- **Zod validation** — Use Zod v4 schemas for all API input validation
+- **Naming**: Files = camelCase/kebab-case, components = PascalCase, constants = UPPER_SNAKE
 
 ---
 
@@ -148,91 +216,97 @@ Current test status: **368+ unit tests** covering:
 
 ```
 src/                        # TypeScript (.ts / .tsx)
-├── app/                    # Next.js App Router
-│   ├── (dashboard)/        # Dashboard pages (.tsx)
-│   ├── api/                # API routes (.ts)
+├── app/                    # Next.js 16 App Router
+│   ├── (dashboard)/        # Dashboard pages (23 sections)
+│   ├── api/                # API routes (51 directories)
 │   └── login/              # Auth pages (.tsx)
-├── domain/                 # Domain types and response helpers (.ts)
+├── domain/                 # Policy engine (policyEngine, comboResolver, costRules, etc.)
 ├── lib/                    # Core business logic (.ts)
-│   ├── db/                 # SQLite database layer
-│   ├── oauth/              # OAuth services per provider
-│   ├── cacheLayer.ts       # LRU cache
-│   ├── semanticCache.ts    # Semantic response cache
-│   ├── idempotencyLayer.ts # Request deduplication
-│   └── localDb.ts          # Settings facade (LowDB for config, SQLite for domain data)
+│   ├── a2a/                # Agent-to-Agent v0.3 protocol server
+│   ├── acp/                # Agent Communication Protocol registry
+│   ├── compliance/         # Compliance policy engine
+│   ├── db/                 # SQLite database layer (21 modules + 16 migrations)
+│   ├── memory/             # Persistent conversational memory
+│   ├── oauth/              # OAuth providers, services, and utilities
+│   ├── skills/             # Extensible skill framework
+│   ├── usage/              # Usage tracking and cost calculation
+│   └── localDb.ts          # Re-export layer only — never add logic here
+├── middleware/              # Request middleware (promptInjectionGuard)
+├── mitm/                   # MITM proxy (cert, DNS, target routing)
 ├── shared/
 │   ├── components/         # React components (.tsx)
-│   ├── middleware/          # Correlation IDs, etc.
-│   ├── utils/              # Circuit breaker, sanitizer, etc.
-│   └── validation/         # Zod schemas
-└── sse/                    # SSE chat handlers (.ts)
+│   ├── constants/          # Provider definitions (177), MCP scopes, 14 routing strategies
+│   ├── utils/              # Circuit breaker, sanitizer, auth helpers
+│   └── validation/         # Zod v4 schemas
+└── sse/                    # SSE proxy pipeline
 
-open-sse/                   # @omniroute/open-sse workspace (JavaScript)
-├── handlers/               # chatCore.js — main request handler
-├── services/               # Rate limit, fallback
-├── translators/            # Format converters (OpenAI ↔ Claude ↔ Gemini)
-└── utils/                  # Progress tracker, stream helpers
+open-sse/                   # @omniroute/open-sse workspace
+├── executors/              # 14 provider-specific request executors
+├── handlers/               # 11 request handlers (chat, responses, embeddings, images, etc.)
+├── mcp-server/             # MCP server (25 tools, 3 transports, 10 scopes)
+├── services/               # 36+ services (combo, autoCombo, rateLimitManager, etc.)
+├── translator/             # Format translators (OpenAI ↔ Claude ↔ Gemini ↔ Responses ↔ Ollama)
+├── transformer/            # Responses API transformer
+└── utils/                  # 22 utility modules (stream, TLS, proxy, logging)
+
+electron/                   # Electron desktop app (cross-platform)
 
 tests/
-├── unit/                   # Node.js test runner (.test.mjs)
-└── e2e/                    # Playwright tests
+├── unit/                   # Node.js test runner (1,574 test files)
+├── integration/            # Integration tests
+├── e2e/                    # Playwright tests
+├── security/               # Security tests
+├── translator/             # Translator-specific tests
+└── load/                   # Load tests
 
-docs/                       # Documentation
-├── USER_GUIDE.md           # Provider setup, CLI integration
-├── API_REFERENCE.md        # All endpoints
-├── TROUBLESHOOTING.md      # Common issues
-├── ARCHITECTURE.md         # System architecture
-└── adr/                    # Architecture Decision Records
+docs/
+├── adr/                     # Architecture Decision Records
+├── architecture/            # System architecture & resilience
+├── comparison/              # OmniRoute vs alternatives
+├── compression/             # Compression guides & rules
+├── dev/                     # Development guides
+├── diagrams/                # Architecture diagrams
+├── frameworks/              # MCP, A2A, OpenCode, Memory, Skills
+├── guides/                  # User guide, Docker, setup, troubleshooting
+├── i18n/                    # Internationalized README translations
+├── marketing/               # Marketing materials
+├── ops/                     # Deployment, proxy, coverage, releases
+├── providers/               # Provider-specific docs
+├── reference/               # API reference, env vars, CLI tools, free tiers
+├── releases/                # Release notes
+├── routing/                 # Auto-combo engine, reasoning replay
+├── screenshots/             # Dashboard screenshots
+├── security/                # Guardrails, compliance, stealth, tokens
+└── specs/                   # Design specs
 ```
 
 ---
 
 ## Adding a New Provider
 
-### Step 1: OAuth Service (if using OAuth)
+### Step 1: Register Provider Constants
 
-Create `src/lib/oauth/services/your-provider.ts` extending `OAuthService`:
+Add to `src/shared/constants/providers.ts` — Zod-validated at module load.
 
-```typescript
-import { OAuthService } from "../OAuthService";
+### Step 2: Add Executor (if custom logic needed)
 
-export class YourProviderService extends OAuthService {
-  constructor() {
-    super({
-      name: "your-provider",
-      authUrl: "https://provider.com/oauth/authorize",
-      tokenUrl: "https://provider.com/oauth/token",
-      clientId: "...",
-      scopes: ["..."],
-    });
-  }
-}
-```
+Create executor in `open-sse/executors/your-provider.ts` extending the base executor.
 
-### Step 2: Register Provider
+### Step 3: Add Translator (if non-OpenAI format)
 
-Add to `src/lib/oauth/providers.ts`:
+Create request/response translators in `open-sse/translator/`.
 
-```typescript
-import { YourProviderService } from "./services/your-provider";
-// Add to the providers map
-```
+### Step 4: Add OAuth Config (if OAuth-based)
 
-### Step 3: Add Constants
+Add OAuth credentials in `src/lib/oauth/constants/oauth.ts` and service in `src/lib/oauth/services/`.
 
-Add provider constants in `src/lib/providerConstants.ts`:
+If the upstream provider distributes a public OAuth client_id/secret or Firebase Web API key inside its public CLI / browser bundle, **do not** embed it as a string literal. Use `resolvePublicCred()` from `open-sse/utils/publicCreds.ts` and add a masked byte entry to `EMBEDDED_DEFAULTS`. The full mandatory workflow is documented in [`docs/security/PUBLIC_CREDS.md`](./docs/security/PUBLIC_CREDS.md).
 
-- Provider prefix (e.g., `yp/`)
-- Default models
-- Pricing info
+Inside handlers/executors, error messages reaching the client must go through `buildErrorBody()` / `sanitizeErrorMessage()` from `open-sse/utils/error.ts` — never put raw `err.stack` or `err.message` in a Response body. See [`docs/security/ERROR_SANITIZATION.md`](./docs/security/ERROR_SANITIZATION.md).
 
-### Step 4: Add Translator (if non-OpenAI format)
+### Step 5: Register Models
 
-Create translator in `open-sse/translators/` if the provider uses a custom API format.
-
-### Step 5: Add Timeout
-
-Add request timeout configuration in `src/shared/utils/requestTimeout.ts`.
+Add model definitions in `open-sse/config/providerRegistry.ts`.
 
 ### Step 6: Add Tests
 
@@ -251,23 +325,33 @@ Write unit tests in `tests/unit/` covering at minimum:
 - [ ] Build succeeds (`npm run build`)
 - [ ] TypeScript types added for new public functions and interfaces
 - [ ] No hardcoded secrets or fallback values
+- [ ] Public upstream credentials embedded via `resolvePublicCred()` (see [`docs/security/PUBLIC_CREDS.md`](./docs/security/PUBLIC_CREDS.md)), never as literals
+- [ ] Error responses route through `buildErrorBody()` / `sanitizeErrorMessage()` — no raw stack traces in response bodies (see [`docs/security/ERROR_SANITIZATION.md`](./docs/security/ERROR_SANITIZATION.md))
+- [ ] Shell commands (`exec` / `spawn`) pass runtime values via `env`, not via string interpolation
+- [ ] All inputs validated with Zod schemas
 - [ ] CHANGELOG updated (if user-facing change)
 - [ ] Documentation updated (if applicable)
+- [ ] No new CodeQL / Secret-Scanning alerts opened, or each one dismissed with technical justification referencing the relevant `docs/security/` doc
+- [ ] Routes that spawn child processes (`/api/mcp/`, `/api/cli-tools/runtime/`) classified as `isLocalOnlyPath()` in `src/server/authz/routeGuard.ts` — see [Hard Rule #15](docs/security/ROUTE_GUARD_TIERS.md)
+- [ ] No `Co-Authored-By` trailers in commit messages — commits must appear solely under the repository owner's Git identity (Hard Rule #16)
 
 ---
 
 ## Releasing
 
-When a new GitHub Release is created (e.g. `v0.4.0`), the package is **automatically published to npm** via GitHub Actions:
+Releases are managed via the `/generate-release` workflow. When a new GitHub Release is created, the package is **automatically published to npm** via GitHub Actions.
 
-```bash
-gh release create v0.4.0 --title "v0.4.0" --generate-notes
-```
+For VPS deploys, use `npm run build:release` (not `npm run build`) — it performs a clean
+rebuild, assembles the bundle into `dist/`, and writes the `dist/BUILD_SHA` sentinel.
+Then use the `/deploy-vps-*-cc` skills which rsync `dist/` to the remote `app/` directory.
 
 ---
 
 ## Getting Help
 
-- **Architecture**: See [`docs/ARCHITECTURE.md`](docs/ARCHITECTURE.md)
+- **Architecture**: See [`docs/architecture/ARCHITECTURE.md`](docs/architecture/ARCHITECTURE.md)
+- **API Reference**: See [`docs/reference/API_REFERENCE.md`](docs/reference/API_REFERENCE.md)
+- **Security docs**: [`docs/security/CLI_TOKEN.md`](docs/security/CLI_TOKEN.md), [`docs/security/ROUTE_GUARD_TIERS.md`](docs/security/ROUTE_GUARD_TIERS.md), [`docs/security/ERROR_SANITIZATION.md`](docs/security/ERROR_SANITIZATION.md), [`docs/security/PUBLIC_CREDS.md`](docs/security/PUBLIC_CREDS.md)
+- **Ops docs**: [`docs/ops/SQLITE_RUNTIME.md`](docs/ops/SQLITE_RUNTIME.md)
 - **Issues**: [github.com/diegosouzapw/OmniRoute/issues](https://github.com/diegosouzapw/OmniRoute/issues)
 - **ADRs**: See `docs/adr/` for architectural decision records

DESING.md

@@ -0,0 +1,254 @@
+# OmniRoute — Design System & Visual Identity
+
+> **Status:** analysis + standardization plan (no code applied yet — this doc is the spec to approve before implementation).
+> **Date:** 2026-06-16 · **Scope:** unify the OmniRoute dashboard (`src/`) with the marketing site (`_mono_repo/omnirouteSite/`) into **one visual identity** — same graph-paper grid background, same color tokens, standardized components.
+
+---
+
+## 1. Purpose
+
+The marketing site (`viral.omniroute.online`, `why.omniroute.online`, `omniroute.online`) and the product dashboard should look like **one product**. The site already borrowed its palette from the dashboard — its `css/tokens.css` even says _"Palette mirrors the OmniRoute dashboard (src/app/globals.css)"_. So the two are already ~80% aligned at the color level. What's missing on the dashboard:
+
+1. The **graph-paper grid wallpaper** the site uses on every page.
+2. A handful of **shared design tokens** the site has but the dashboard lacks (radius scale, brand gradient, `surface-2`, mono font).
+3. **Component-level consistency** — a number of dashboard components bypass the theme tokens with hardcoded hex/rgba.
+
+This document is the analysis and the plan. **Nothing is changed until approved.**
+
+---
+
+## 2. Principles
+
+- **Single source of truth = `src/app/globals.css`.** The site mirrors the dashboard, never the other way around. New tokens land in `globals.css` first.
+- **Tokens, never literals.** Components consume semantic tokens (`bg-surface`, `text-primary`, `border-border`), never raw `#hex`.
+- **Subtle, not loud.** The grid is a faint wallpaper that sits behind content — it must never reduce text contrast or fight the UI.
+- **Theme-aware.** Everything works in both `.dark` (default-ish, the product's signature look) and light.
+- **Surgical rollout.** Ship the grid + tokens first (low risk, high visibility), then component cleanups in waves.
+
+---
+
+## 3. Current state — what's already aligned vs. what's not
+
+### 3.1 Colors — already unified ✅
+
+Every brand color and surface already matches the site **by value** (only the names differ — dashboard prefixes with `--color-`). Verified in `src/app/globals.css:30-128`:
+
+| Concept                    | Site token (`tokens.css`)                   | Dashboard token (`globals.css`) | Match        |
+| -------------------------- | ------------------------------------------- | ------------------------------- | ------------ |
+| primary                    | `--primary #e54d5e`                         | `--color-primary #e54d5e`       | ✅           |
+| primary-hover              | `--primary-hover #c93d4e`                   | `--color-primary-hover #c93d4e` | ✅           |
+| accent                     | `--accent #6366f1`                          | `--color-accent #6366f1`        | ✅           |
+| accent-2                   | `--accent-2 #8b5cf6`                        | `--color-accent-hover #8b5cf6`  | ✅ (renamed) |
+| accent-3                   | `--accent-3 #a855f7`                        | `--color-accent-light #a855f7`  | ✅ (renamed) |
+| success / warning / error  | `#22c55e / #f59e0b / #ef4444`               | identical                       | ✅           |
+| traffic lights             | `#ff5f56 / #ffbd2e / #27c93f`               | identical                       | ✅           |
+| dark bg / surface / border | `#0b0e14 / #161b22 / rgba(255,255,255,.08)` | identical                       | ✅           |
+| light bg / surface / text  | `#f9f9fb / #fff / #1a1a2e`                  | identical                       | ✅           |
+
+**Conclusion:** there is no color migration to do. The identity is already shared; we are _finishing_ it, not rebuilding it.
+
+### 3.2 Gaps — what the dashboard is missing
+
+| Gap                     | Site has                                                                       | Dashboard                                                  | Action                 |
+| ----------------------- | ------------------------------------------------------------------------------ | ---------------------------------------------------------- | ---------------------- |
+| **Grid wallpaper**      | `body::before` graph-paper, `--grid-line`, `--grid-size 46px`, `--section-alt` | none (flat `--color-bg`)                                   | **Part A**             |
+| **Radius scale**        | `--radius 14px`, `--radius-sm 9px`                                             | none — primitives use ad-hoc `rounded-md/lg/xl` (6/8/12px) | **Part B**             |
+| **Brand gradient**      | `--grad-brand 135deg primary→accent-3`                                         | none — only a one-off `.bg-hero-gradient`                  | **Part B**             |
+| **Nested surface**      | `--surface-2 #1c2230`                                                          | none                                                       | **Part B**             |
+| **Mono font**           | `--font-mono` (ui-monospace stack)                                             | none (code/terminal areas have no token)                   | **Part B**             |
+| **`text-muted` (dark)** | `#8b8b9e`                                                                      | `#a1a1aa` (zinc-400)                                       | reconcile — **Part B** |
+
+### 3.3 Theming mechanics (so we don't break anything)
+
+- **Tailwind v4, CSS-first** (no `tailwind.config.*`). Tokens are defined in `:root`/`.dark` and exposed to utilities via `@theme inline` (`globals.css:130-179`).
+- **Dark via `.dark` class** on `<html>` (`@custom-variant dark` at `globals.css:22`), toggled by a custom Zustand store (`src/store/themeStore.ts`), default theme = `system` (`src/shared/constants/appConfig.ts:11`). The site uses `html[data-theme="light"]` instead — **the mechanisms differ but never meet** (separate origins), so no conflict. We keep the dashboard's `.dark` mechanism.
+- **Runtime primary override** exists (`themeStore.ts:85-97`, presets in `COLOR_THEMES`) — users can swap `--color-primary`. Any new token (gradient, etc.) that references `--color-primary` will inherit those overrides for free. ✅
+
+---
+
+## 4. Part A — The graph-paper grid background (headline ask)
+
+### 4.1 What it is
+
+The exact recipe from the site (`_mono_repo/omnirouteSite/css/base.css`): a **fixed, full-viewport pseudo-element** painting two 1px line gradients, sitting at `z-index:-1` behind all content.
+
+```css
+body::before {
+  content: "";
+  position: fixed;
+  inset: 0;
+  z-index: -1;
+  pointer-events: none;
+  background-image:
+    linear-gradient(to right, var(--grid-line) 1px, transparent 1px),
+    linear-gradient(to bottom, var(--grid-line) 1px, transparent 1px);
+  background-size: var(--grid-size) var(--grid-size);
+}
+```
+
+**Why this works even though `body` has an opaque `background-color`:** a `::before` with `z-index:-1` paints _above_ the element's own background but _below_ its in-flow content. So `--color-bg` is the base fill, the grid is layered on top of it, and the app renders above the grid.
+
+### 4.2 Precedent already in the codebase
+
+`src/app/landing/page.tsx:16-26` **already implements this same grid per-page** — but with **red** lines (`#E54D5E`, opacity `0.06`) at **50px**, plus animated orbs. So the pattern is proven in the product; we are promoting it to a **global, theme-aware** wallpaper and (optionally) retiring the duplicate.
+
+### 4.3 Tokens to add (in `globals.css`)
+
+```css
+:root {
+  /* light */
+  --grid-line: rgba(0, 0, 0, 0.045);
+  --grid-size: 46px;
+  --section-alt: rgba(0, 0, 0, 0.022);
+}
+.dark {
+  /* dark */
+  --grid-line: rgba(255, 255, 255, 0.035);
+  --section-alt: rgba(255, 255, 255, 0.018);
+}
+```
+
+### 4.4 The single blocker
+
+The grid is global by construction (it covers the panel, `auth`/`login`, error pages — every route — at once). Exactly **one** element hides it inside the panel:
+
+- `src/shared/components/layouts/DashboardLayout.tsx:62` — the outer wrapper paints an opaque `bg-bg`:
+
+  ```jsx
+  <div className="flex h-dvh min-h-0 w-full overflow-hidden bg-bg">
+  ```
+
+  Everything below it is already transparent — `<main>` (`:93`), the scroll container (`:102`), the `max-w-7xl` inner (`:103`). So **removing `bg-bg` from this one line** lets the body grid show through the entire content area (the body's `--color-bg` remains the base fill underneath the grid).
+
+  ```diff
+  - <div className="flex h-dvh min-h-0 w-full overflow-hidden bg-bg">
+  + <div className="flex h-dvh min-h-0 w-full overflow-hidden">
+  ```
+
+### 4.5 Chrome interaction (sidebar / header)
+
+- `Header` (`src/shared/components/Header.tsx:207`, `bg-bg`) and `Sidebar` (`src/shared/components/Sidebar.tsx:430`, `bg-sidebar`) stay **opaque** → the grid shows in the **content area only**, with solid chrome framing it. This is the recommended, calm default and matches how the site separates chrome from canvas.
+- _Optional vibrancy variant:_ make the header translucent (`bg-bg/80 backdrop-blur`) so the grid runs behind it. A `.bg-vibrancy` helper already exists (`globals.css:370`). **Decision D3 below.**
+
+### 4.6 Login / auth / error pages
+
+These render directly under `<body>` (no panel chrome) and their page wrappers are mostly transparent — the global grid appears behind them automatically. One exception: `src/app/login/page.tsx:124,139` uses opaque `bg-bg` wrappers; soften the same way if we want the grid there too (minor, **D4**).
+
+### 4.7 Landing page
+
+`landing/page.tsx` keeps its richer animated background (orbs + vignette). Options: (a) leave it as-is (its own splash identity), or (b) align its grid to the global tokens (46px, neutral lines) for consistency. **Recommend (a)** — it's a marketing splash, not a panel screen. **Decision D5.**
+
+---
+
+## 5. Part B — Token unification
+
+Add to `globals.css` (`:root` + `@theme inline`) so the dashboard gains the site's missing tokens. None of these change existing colors; they add the _missing_ primitives.
+
+```css
+:root {
+  --surface-2: #f5f5fa; /* light: nested panels */
+  --radius: 14px;
+  --radius-sm: 9px;
+  --grad-brand: linear-gradient(135deg, var(--color-primary), var(--color-accent-light));
+  --font-mono: ui-monospace, "JetBrains Mono", "Fira Code", "SF Mono", monospace;
+}
+.dark {
+  --surface-2: #1c2230;
+}
+
+@theme inline {
+  --color-surface-2: var(--surface-2); /* enables bg-surface-2 */
+  --radius-lg: var(--radius); /* enables rounded-lg = 14px */
+  --radius-md: var(--radius-sm); /* enables rounded-md = 9px */
+  --font-mono: var(--font-mono); /* enables font-mono */
+}
+```
+
+| Token                      | Why                                                             | Consumers                                                   |
+| -------------------------- | --------------------------------------------------------------- | ----------------------------------------------------------- |
+| `--radius` / `--radius-sm` | One radius scale (14/9) instead of 6/8/12 ad-hoc                | Button, Card, Modal, Input, Select                          |
+| `--grad-brand`             | Brand gradient for primary CTAs (red→violet), matching the site | Button `primary`, hero/CTA surfaces                         |
+| `--surface-2`              | Nested panels / table headers / inset rows                      | Card.Section, DataTable header, inputs                      |
+| `--font-mono`              | Code blocks, terminal, IDs, endpoints                           | ConsoleLogViewer, code snippets, `localhost:20128/v1` chips |
+| `--text-muted` reconcile   | Pick one value site↔panel                                       | global                                                      |
+
+**Decision D2 (text-muted):** site `#8b8b9e` vs dashboard `#a1a1aa`. Recommend keeping the **dashboard's `#a1a1aa`** (it's the live product, slightly higher contrast) and updating the _site_ to match. Low priority, cosmetic.
+
+---
+
+## 6. Part C — Component standardization
+
+The component layer is **custom** (no shadcn/Radix), Tailwind v4, semantic tokens **mostly** adopted (`bg-surface`, `border-white/10`, `ring-primary`) — good adoption (195 files import the shared barrel). The work is removing the **bypasses**. Home: `src/shared/components/`.
+
+Ranked by impact × reach:
+
+| #   | Item                                       | File(s)                                                                                                                                                                                       | Problem → Target                                                                                                                                                                                                                                 |
+| --- | ------------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
+| C1  | **Radius alignment**                       | `Button.tsx:14-18`, `Card.tsx:39`, `Modal.tsx`, `Input.tsx`, `Select.tsx`                                                                                                                     | mixed 6/8/12px → repoint to `--radius`/`--radius-sm` (14/9)                                                                                                                                                                                      |
+| C2  | **Button gradient + `accent` variant**     | `Button.tsx:5-12`                                                                                                                                                                             | primary is flat red→red (`from-primary to-primary-hover`); align to `--grad-brand` (red→violet) and add the missing `accent` variant (indigo `#6366f1` is unused by buttons) — **highest visibility, ~195 importers**. **Decision D1.**          |
+| C3  | **Tables**                                 | `DataTable.tsx:122-176`, `logTableStyles.ts`, `globals.css:405-414` (Ant remnants)                                                                                                            | `DataTable` is 100% inline hardcoded rgba + references non-existent vars (`--text-secondary`, `--bg-table-header`); migrate to tokens, retire the 2 divergent table styles. Tables are everywhere (providers/connections/logs) — worst offender. |
+| C4  | **Centralize status colors**               | `flow/edgeStyles.ts:7-12`, `TokenHealthBadge.tsx:14-19`, `DegradationBadge.tsx`, `ProviderCascadeNode.tsx`, `Badge.tsx`, +5 `statusColor` helpers                                             | 6+ copies of the same `#22c55e/#f59e0b/#ef4444` hex; create one `statusColors` module driven off `--color-success/warning/error`. Critical for circuit-breaker / cooldown / lockout badges to read consistently.                                 |
+| C5  | **Card border**                            | `Card.tsx:39`                                                                                                                                                                                 | uses `border-white/5`; brand border is `/8` → align                                                                                                                                                                                              |
+| C6  | **Focus ring reconcile**                   | `globals.css:183` vs component `ring-primary/30`                                                                                                                                              | global `:focus-visible` is indigo (`--color-accent`), components are red (`ring-primary`) — pick one (recommend **accent/indigo** globally, it reads as the "interactive" color)                                                                 |
+| C7  | **Add `Checkbox` + `Textarea` primitives** | currently raw `<input>`/`<textarea>` with inline `accentColor:#6366f1` (e.g. `ColumnToggle.tsx:91`)                                                                                           | create token-driven primitives                                                                                                                                                                                                                   |
+| C8  | **Hardcoded-hex sweep**                    | `ConsoleLogViewer.tsx:240` (`#161b22`/`#30363d`), `ComboLiveStudio.tsx:306` (`#6366f120`), Modal traffic dots `Modal.tsx:149-159`, ~14 chart/component files with literal `#6366f1`/`#a855f7` | replace literals with `bg-surface`/`border-border`/`text-accent` etc.                                                                                                                                                                            |
+| C9  | **`cn()` → clsx + tailwind-merge**         | `src/shared/utils/cn.ts`                                                                                                                                                                      | current `cn` just joins; conflicting classes stack (a `className="rounded-2xl"` override won't replace a primitive's `rounded-lg`). Needed for C1 overrides to behave.                                                                           |
+
+**Already on-brand (token-driven, only need radius):** `Badge`, `Toggle`, `SegmentedControl`, `Input`, `Select`.
+
+---
+
+## 7. Rollout plan (phased, each phase shippable + testable)
+
+- **Phase 1 — Grid + tokens (low risk, high visibility).**
+  1. Add grid + identity tokens to `globals.css` (Part A §4.3, Part B §5).
+  2. Add `body::before` grid.
+  3. Remove `bg-bg` from `DashboardLayout.tsx:62`.
+  4. Verify across themes + key screens (dashboard, providers, logs, login, an error page). Confirm contrast unchanged.
+     → _Delivers the headline ask. Reversible in one commit._
+
+- **Phase 2 — Primitives radius + Button (C1, C2, C5, C9).** The visible "feel" pass. `cn()` upgrade first so overrides behave.
+
+- **Phase 3 — Tables + status colors (C3, C4).** The largest consistency win; touch the data-heavy screens.
+
+- **Phase 4 — Cleanup (C6, C7, C8).** Focus ring, new primitives, hardcoded-hex sweep.
+
+Each phase: `npm run lint` + `npm run typecheck:core` + a visual pass. Per repo rule, production-code changes ship with tests where applicable (token/CSS changes are visual — validated by screenshots; component API changes get unit coverage).
+
+---
+
+## 8. Open decisions (need your call before/while implementing)
+
+- **D1 — Button primary look.** Keep the current **red→red** gradient, or switch the product's primary buttons to the **red→violet `--grad-brand`** (matches the site CTAs)? _(Affects every primary button.)_ Recommend: **red→violet**, with `--grad-brand`.
+- **D2 — Grid line color.** **Neutral** lines (site style: faint white/black, `rgba(255,255,255,0.035)`) — calm, content-first — **or** the landing's **brand-red** lines? Recommend: **neutral** (matches the site's interior pages; red is louder and can tint readability). Size **46px** (site) to retire the landing's 50px drift.
+- **D3 — Chrome vibrancy.** Sidebar/header stay **solid** (grid in content area only), or go **translucent** so the grid runs behind them? Recommend: **solid** (calmer; less risk).
+- **D4 — Auth/login grid.** Soften `login/page.tsx` wrappers so the grid shows there too? Recommend: **yes** (cheap, more cohesive).
+- **D5 — Landing page.** Leave its animated splash bg as-is, or align it to the global grid? Recommend: **leave as-is**.
+- **D6 — Radius value.** Adopt **14/9** everywhere (bigger, softer, site-matching) — confirm you want this product-wide shift. Recommend: **yes**, it's the single biggest "one identity" signal.
+- **D7 — Scope of first PR.** Ship **Phase 1 only** first (grid + tokens), then iterate? Recommend: **yes** — validate the wallpaper live before the component waves.
+
+---
+
+## 9. Out of scope / risks
+
+- **No palette change** — colors already match; we only add missing tokens. Zero risk of recoloring the product.
+- **No theme-engine change** — keep `.dark` + Zustand store; don't migrate to `next-themes` or to the site's `data-theme`.
+- **Radius shift is broad** (D6) — it touches every card/button/input; that's the point, but it's the one change worth eyeballing on busy screens (tables, modals) before merge.
+- **Tables (C3)** carry the most hardcoded styling and the highest regression surface — isolate in its own PR with before/after screenshots.
+- **Worktree isolation (repo hard-rule #19):** implementation runs in a dedicated worktree on a branch cut from the confirmed base (likely `release/v3.8.28`), never on the shared checkout. This doc is the only artifact written to the working tree so far.
+
+---
+
+## 10. Reference index
+
+| Area                       | Path                                                                                                                                     |
+| -------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------- |
+| Dashboard tokens           | `src/app/globals.css:30-179` (`:root`, `.dark`, `@theme inline`), `body` `:206`                                                          |
+| Theme store                | `src/store/themeStore.ts`, `src/shared/components/ThemeProvider.tsx`, `src/shared/constants/appConfig.ts:9-11`                           |
+| Panel shell (grid blocker) | `src/shared/components/layouts/DashboardLayout.tsx:62`                                                                                   |
+| Chrome                     | `src/shared/components/Header.tsx:207`, `src/shared/components/Sidebar.tsx:430`                                                          |
+| Grid precedent             | `src/app/landing/page.tsx:16-26`                                                                                                         |
+| Primitives                 | `src/shared/components/{Button,Card,Input,Select,Badge,Modal,Toggle,SegmentedControl,Loading,Tooltip,DataTable}.tsx`, barrel `index.tsx` |
+| Status-color sources       | `flow/edgeStyles.ts`, `TokenHealthBadge.tsx`, `DegradationBadge.tsx`, `logTableStyles.ts`                                                |
+| `cn` util                  | `src/shared/utils/cn.ts`                                                                                                                 |
+| Site reference             | `_mono_repo/omnirouteSite/css/tokens.css`, `css/base.css` (grid `body::before`)                                                          |

Dockerfile

@@ -1,14 +1,68 @@
-FROM node:22-bookworm-slim AS builder
+# ── Common base with runtime deps ──────────────────────────────────────────
+FROM node:24-trixie-slim AS base
 WORKDIR /app
 
+RUN --mount=type=cache,target=/var/cache/apt,sharing=shared \
+  --mount=type=cache,target=/var/lib/apt/lists,sharing=shared \
+  apt-get update \
+  && apt-get install -y --no-install-recommends libsecret-1-0 ca-certificates \
+  && rm -rf /var/lib/apt/lists/*
+
+# ── Builder ────────────────────────────────────────────────────────────────
+FROM base AS builder
+
+# Build tools for native module compilation
+# apt-get update needed here because base's rm -rf clears the shared cache
+RUN --mount=type=cache,target=/var/cache/apt,sharing=shared \
+  --mount=type=cache,target=/var/lib/apt/lists,sharing=shared \
+  apt-get update \
+  && apt-get install -y --no-install-recommends python3 make g++ \
+  && rm -rf /var/lib/apt/lists/*
+
 COPY package*.json ./
-RUN if [ -f package-lock.json ]; then npm ci --no-audit --no-fund; else npm install --no-audit --no-fund; fi
+COPY scripts/build/postinstall.mjs ./scripts/build/postinstall.mjs
+COPY scripts/build/postinstallSupport.mjs ./scripts/build/postinstallSupport.mjs
+COPY scripts/build/native-binary-compat.mjs ./scripts/build/native-binary-compat.mjs
+ENV NPM_CONFIG_LEGACY_PEER_DEPS=true
+# --ignore-scripts blocks broad dependency install/postinstall hooks, closing
+# the supply-chain attack surface where a transitive dep can run arbitrary code
+# at install time. better-sqlite3 still needs a native binding for the target
+# platform, so rebuild and smoke-test only that known runtime dependency below.
+#
+# We REQUIRE a committed package-lock.json so resolved dependency versions
+# are reproducible.
+RUN test -f package-lock.json \
+  || (echo "package-lock.json is required for reproducible Docker builds" >&2 && exit 1)
+RUN --mount=type=cache,target=/root/.npm \
+  npm ci --no-audit --no-fund --legacy-peer-deps --ignore-scripts \
+  && npm rebuild better-sqlite3 \
+  && node -e "require('better-sqlite3')(':memory:').close()"
+
+# Build with webpack (stable). Turbopack hit a non-recoverable internal panic on this
+# Next.js version during the v3.8.27 release build — TurbopackInternalError "entered
+# unreachable code: there must be a path to a root" in ImportTracer::get_traces, on both
+# linux/amd64 and linux/arm64. Webpack is the proven engine (build:release / VPS / CI Build
+# all green). Re-enable Turbopack (=1) once the upstream tracer bug is fixed.
+# See docs/ops/QUALITY_GATE_PLAYBOOK.md Parte 6.
+ENV OMNIROUTE_USE_TURBOPACK=0
+
+# Raise the V8 heap ceiling for the build. The webpack production optimization
+# pass (forced above since Turbopack panics) needs more than V8's default ceiling
+# (~2 GB) for a codebase this size; a memory-constrained Docker build otherwise
+# dies with "FATAL ERROR: ... JavaScript heap out of memory" at `[builder] npm run
+# build` (#4076). NODE_OPTIONS propagates to the spawned `next build` child
+# (build-next-isolated.mjs → resolveNextBuildEnv spreads process.env). Build-only;
+# the runtime heap is set separately on the runner stage (OMNIROUTE_MEMORY_MB).
+# Override for hosts with more/less RAM: `--build-arg OMNIROUTE_BUILD_MEMORY_MB=6144`.
+ARG OMNIROUTE_BUILD_MEMORY_MB=4096
+ENV NODE_OPTIONS="--max-old-space-size=${OMNIROUTE_BUILD_MEMORY_MB}"
 
 COPY . ./
-RUN mkdir -p /app/data && npm run build
+RUN --mount=type=cache,target=/app/.build/next/cache \
+  mkdir -p /app/data && npm run build
 
-FROM node:22-bookworm-slim AS runner-base
-WORKDIR /app
+# ── Runner base ────────────────────────────────────────────────────────────
+FROM base AS runner-base
 
 LABEL org.opencontainers.image.title="omniroute" \
   org.opencontainers.image.description="Unified AI proxy — route any LLM through one endpoint" \
@@ -19,30 +73,113 @@ LABEL org.opencontainers.image.title="omniroute" \
 ENV NODE_ENV=production
 ENV PORT=20128
 ENV HOSTNAME=0.0.0.0
+ENV OMNIROUTE_MEMORY_MB=1024
+ENV NODE_OPTIONS="--max-old-space-size=${OMNIROUTE_MEMORY_MB}"
 
 # Data directory inside Docker — must match the volume mount in docker-compose.yml
 ENV DATA_DIR=/app/data
 RUN mkdir -p /app/data
 
-COPY --from=builder /app/public ./public
-COPY --from=builder /app/.next/static ./.next/static
-COPY --from=builder /app/.next/standalone ./
+# `npm run build` (build-next-isolated → assembleStandalone) bundles ALL runtime
+# files into .build/next/standalone/ — .next, node_modules, migrations, scripts,
+# docs, and the previously hand-COPY'd modules below (@swc/helpers, pino-*, split2,
+# migrations). assembleStandalone copies them straight from the builder's
+# node_modules, so they are present regardless of NFT/Turbopack trace behaviour.
+# The old per-module overrides were therefore pure duplication and were removed
+# (build-output-isolation cleanup). See scripts/build/assembleStandalone.mjs
+# (EXTRA_MODULE_ENTRIES) for the single source of truth.
+COPY --from=builder /app/.build/next/standalone ./
+# better-sqlite3 is the one exception still copied explicitly: assembleStandalone
+# only syncs its native build/ dir; the JS wrapper (lib/, package.json) is left to
+# Next.js tracing. bootstrap-env requires SQLite BEFORE the standalone server
+# starts, so guarantee the complete package independent of trace behaviour.
+COPY --from=builder /app/node_modules/better-sqlite3 ./node_modules/better-sqlite3
+# migrations land at <standalone>/migrations via assembleStandalone; point the runtime at them.
+ENV OMNIROUTE_MIGRATIONS_DIR=/app/migrations
+
+# Docker healthcheck script — not traced by Next.js standalone output, so copy
+# it explicitly. The HEALTHCHECK CMD references it as `node healthcheck.mjs`.
+COPY --from=builder /app/scripts/dev/healthcheck.mjs ./healthcheck.mjs
+
+# Hand /app over to the baked-in `node` non-root user (UID/GID 1000) so the
+# runtime process never holds root privileges. The chown happens after all
+# COPYs so it covers files originally owned by root in the builder stage.
+RUN chown -R node:node /app
 
 EXPOSE 20128
 
-HEALTHCHECK --interval=30s --timeout=5s --start-period=15s --retries=3 \
-  CMD node -e "fetch('http://127.0.0.1:20128/api/settings').then(r=>{if(!r.ok)throw r.status}).catch(()=>process.exit(1))"
+# Drop to non-root before ENTRYPOINT/CMD so every derived stage (runner-cli,
+# runner-web) also runs as a non-root user unless they explicitly switch back.
+USER node
 
-CMD ["node", "server.js"]
+# Warns if the mounted data volume has wrong ownership
+COPY --chmod=755 scripts/check-permissions.sh /tmp/check-permissions.sh
+ENTRYPOINT ["/tmp/check-permissions.sh"]
+
+HEALTHCHECK --interval=30s --timeout=5s --start-period=15s --retries=3 \
+  CMD ["node", "healthcheck.mjs"]
+
+CMD ["node", "dev/run-standalone.mjs"]
+
+# ── Runner Web (web-cookie providers: Gemini Web, Claude Turnstile) ───────────
+#
+#  Two image flavors:
+#    runner-base  →  omniroute:VERSION        Lean base (~500 MB). No browsers.
+#    runner-web   →  omniroute:VERSION-web    +Chromium/Playwright (~800 MB).
+#
+#  Use runner-web when you need web-cookie providers (gemini-web, claude-web,
+#  claude-turnstile). For all other providers runner-base is sufficient.
+#
+#  Build:
+#    docker build --target runner-web -t omniroute:web .
+#  Compose:
+#    build:
+#      context: .
+#      target: runner-web
+FROM runner-base AS runner-web
+
+USER root
+
+# Copy playwright and playwright-core from the builder stage.
+# The slim runtime image does not have playwright in node_modules, so npx falls
+# back to a registry download — unreliable on CI runners (exits 127 on failure).
+# Copying from the builder avoids any network access at image-build time and also
+# ensures the same playwright version is available at runtime for web-session providers.
+COPY --from=builder /app/node_modules/playwright-core ./node_modules/playwright-core
+COPY --from=builder /app/node_modules/playwright ./node_modules/playwright
+
+# Install Playwright browser binaries + OS dependencies under root, then hand
+# ownership of the browsers cache to the node user.
+# PLAYWRIGHT_BROWSERS_PATH overrides the default ~/.cache/ms-playwright so the
+# browsers land under /home/node which persists across image layers and is
+# accessible to the non-root runtime user.
+ENV PLAYWRIGHT_BROWSERS_PATH=/home/node/.cache/ms-playwright
+RUN --mount=type=cache,target=/var/cache/apt,sharing=locked \
+  --mount=type=cache,target=/var/lib/apt/lists,sharing=locked \
+  apt-get update \
+  && node node_modules/playwright/cli.js install chromium --with-deps \
+  && chown -R node:node /home/node/.cache \
+  && rm -rf /var/lib/apt/lists/*
+
+USER node
 
 FROM runner-base AS runner-cli
 
+# Drop back to root briefly so we can install system + global npm packages,
+# then return to the `node` non-root user before the CMD inherited from
+# runner-base runs.
+USER root
+
 # Install system dependencies required by openclaw (git+ssh references).
-RUN apt-get update \
-  && apt-get install -y --no-install-recommends git ca-certificates \
+RUN --mount=type=cache,target=/var/cache/apt,sharing=locked \
+  --mount=type=cache,target=/var/lib/apt/lists,sharing=locked \
+  apt-get update \
+  && apt-get install -y --no-install-recommends git ca-certificates docker.io docker-compose \
   && rm -rf /var/lib/apt/lists/* \
   && git config --system url."https://github.com/".insteadOf "ssh://git@github.com/"
 
 # Install CLI tools globally. Separate layer from apt for better cache reuse.
-RUN npm install -g --no-audit --no-fund @openai/codex @anthropic-ai/claude-code droid openclaw@latest
+RUN --mount=type=cache,target=/root/.npm \
+  npm install -g --no-audit --no-fund @openai/codex @anthropic-ai/claude-code droid openclaw@latest
 
+USER node

GEMINI.md

@@ -0,0 +1,50 @@
+# Security and Cleanliness Rules for AI Assistants
+
+> **Scope:** rules for Gemini-based agents. For Claude Code, see `CLAUDE.md`. For other AI assistants, see `AGENTS.md`.
+
+## 1. File Placement & Organization
+
+- **Test Files**: ALL unit tests, integration tests, ecosystem tests, or Vitest files MUST strictly be placed within the `tests/` directory (e.g., `tests/unit/`, `tests/integration/`). NEVER create test files in the project root (`/`).
+- **Scripts and Utilities**: ALL maintenance, debugging, generation, or experimental scripts (`.cjs`, `.mjs`, `.js`, `.ts`) MUST be placed strictly inside one of the `scripts/` subfolders (`build/`, `dev/`, `check/`, `docs/`, `i18n/`, `ad-hoc/`). One-shot or experimental code goes under `scripts/ad-hoc/`. NEVER dump loose scripts in the project root (`/`) or the top-level `scripts/` folder.
+
+**The Project Root MUST ONLY CONTAIN:**
+
+- Configuration files (`vitest.config.ts`, `next.config.mjs`, `eslint.config.mjs`, `tsconfig*.json`, `playwright.config.ts`, `prettier.config.mjs`, `postcss.config.mjs`, `sonar-project.properties`, `fly.toml`, `docker-compose*.yml`, `Dockerfile`)
+- Dependency files (`package.json`, `package-lock.json`)
+- Documentation files (`README.md`, `CHANGELOG.md`, `LICENSE`, `AGENTS.md`, `CLAUDE.md`, `GEMINI.md`, `CONTRIBUTING.md`, `SECURITY.md`, `CODE_OF_CONDUCT.md`, `llm.txt`, `Tuto_Qdrant.md`)
+- CI/CD files and ignore definitions (`.gitignore`, `.dockerignore`, `.npmignore`, `.npmrc`, `.node-version`, `.nvmrc`, `.env.example`)
+
+When creating _any_ validation tests or one-off logic scripts, default to using `scripts/ad-hoc/` or the `tests/unit/` directories according to your goals. Do not pollute the `/` root context.
+
+## 2. Hard Rules (mirror of `CLAUDE.md`)
+
+1. **Never commit secrets or credentials.** Use `.env` (auto-generated from `.env.example`) or a vault. Passwords, OAuth secrets, API keys, and Cookie values must never appear in committed files.
+2. **Never add logic to `src/lib/localDb.ts`.** It is a re-export barrel only.
+3. **Never use `eval()`, `new Function()`, or any implied eval.** ESLint enforces this.
+4. **Never commit directly to `main`.** Use `feat/`, `fix/`, `refactor/`, `docs/`, `test/`, or `chore/` branches.
+5. **Never write raw SQL in routes** — always go through `src/lib/db/` domain modules.
+6. **Never silently swallow errors in SSE streams** — propagate them or abort the stream cleanly.
+7. **Never bypass Husky hooks** (`--no-verify`, `--no-gpg-sign`) without explicit operator approval.
+8. **Always validate inputs with Zod schemas** from `src/shared/validation/schemas.ts`.
+9. **Always include tests when changing production code** (`src/`, `open-sse/`, `electron/`, `bin/`).
+10. **Coverage must stay** ≥ 75 % statements / 75 % lines / 75 % functions / 70 % branches (real measured: ~82 %).
+
+## 3. Codebase navigation
+
+| Task                    | Read this first                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                              |
+| ----------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| Understand the codebase | `docs/architecture/REPOSITORY_MAP.md`                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                        |
+| Architecture overview   | `docs/architecture/ARCHITECTURE.md`                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                          |
+| Engineering reference   | `docs/architecture/CODEBASE_DOCUMENTATION.md`                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                |
+| Add a feature           | `CONTRIBUTING.md` + the matching `docs/<area>.md`                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            |
+| Per-area deep dives     | `docs/frameworks/SKILLS.md`, `docs/frameworks/MEMORY.md`, `docs/frameworks/EVALS.md`, `docs/security/GUARDRAILS.md`, `docs/security/COMPLIANCE.md`, `docs/frameworks/CLOUD_AGENT.md`, `docs/frameworks/MCP-SERVER.md`, `docs/frameworks/A2A-SERVER.md`, `docs/architecture/AUTHZ_GUIDE.md`, `docs/architecture/RESILIENCE_GUIDE.md`, `docs/routing/AUTO-COMBO.md`, `docs/frameworks/WEBHOOKS.md`, `docs/routing/REASONING_REPLAY.md`, `docs/security/STEALTH_GUIDE.md`, `docs/ops/TUNNELS_GUIDE.md`, `docs/guides/ELECTRON_GUIDE.md`, `docs/reference/PROVIDER_REFERENCE.md` |
+| Release flow            | `docs/ops/RELEASE_CHECKLIST.md`                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                              |
+
+## 4. Local development access
+
+The dashboard is reachable at the operator's chosen URL/port (default `http://localhost:20128`). Credentials are operator-specific:
+
+- **Initial admin password** is read from the `INITIAL_PASSWORD` env var on first install (defaults to `CHANGEME` in `.env.example`; rotate immediately after first login).
+- **Local VPS / shared dev environments**: ask the operator for the URL and current credentials — they live in their personal vault, NOT in this repo.
+
+> Any credential observed in a previous version of this file was a non-production demo value; treat it as compromised and do not reuse it.

README.es.md

@@ -1,999 +0,0 @@
-<div align="center">
-  <img src="./docs/screenshots/MainOmniRoute.png" alt="OmniRoute Dashboard" width="800"/>
-  
-  # 🚀 OmniRoute — El Gateway de IA Gratuito
-
-### Nunca dejes de programar. Enrutamiento inteligente hacia **modelos de IA GRATUITOS y económicos** con fallback automático.
-
-_Tu proxy de API universal — un endpoint, 36+ proveedores, cero tiempo de inactividad._
-
-**Chat Completions • Embeddings • Generación de Imágenes • Audio • Reranking • 100% TypeScript**
-
----
-
-### 🤖 Proveedor de IA Gratuito para tus agentes de programación favoritos
-
-_Conecta cualquier IDE o herramienta CLI con IA a través de OmniRoute — gateway de API gratuito para programación ilimitada._
-
-  <table>
-    <tr>
-      <td align="center" width="110">
-        <a href="https://github.com/cline/cline">
-          <img src="./public/providers/openclaw.png" alt="OpenClaw" width="48"/><br/>
-          <b>OpenClaw</b>
-        </a><br/>
-        <sub>⭐ 205K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/HKUDS/nanobot">
-          <img src="./public/providers/nanobot.png" alt="NanoBot" width="48"/><br/>
-          <b>NanoBot</b>
-        </a><br/>
-        <sub>⭐ 20.9K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/sipeed/picoclaw">
-          <img src="./public/providers/picoclaw.jpg" alt="PicoClaw" width="48"/><br/>
-          <b>PicoClaw</b>
-        </a><br/>
-        <sub>⭐ 14.6K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/zeroclaw-labs/zeroclaw">
-          <img src="./public/providers/zeroclaw.png" alt="ZeroClaw" width="48"/><br/>
-          <b>ZeroClaw</b>
-        </a><br/>
-        <sub>⭐ 9.9K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/nearai/ironclaw">
-          <img src="./public/providers/ironclaw.png" alt="IronClaw" width="48"/><br/>
-          <b>IronClaw</b>
-        </a><br/>
-        <sub>⭐ 2.1K</sub>
-      </td>
-    </tr>
-    <tr>
-      <td align="center" width="110">
-        <a href="https://github.com/anomalyco/opencode">
-          <img src="./public/providers/opencode.svg" alt="OpenCode" width="48"/><br/>
-          <b>OpenCode</b>
-        </a><br/>
-        <sub>⭐ 106K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/openai/codex">
-          <img src="./public/providers/codex.png" alt="Codex CLI" width="48"/><br/>
-          <b>Codex CLI</b>
-        </a><br/>
-        <sub>⭐ 60.8K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/anthropics/claude-code">
-          <img src="./public/providers/claude.png" alt="Claude Code" width="48"/><br/>
-          <b>Claude Code</b>
-        </a><br/>
-        <sub>⭐ 67.3K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/google-gemini/gemini-cli">
-          <img src="./public/providers/gemini-cli.png" alt="Gemini CLI" width="48"/><br/>
-          <b>Gemini CLI</b>
-        </a><br/>
-        <sub>⭐ 94.7K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/Kilo-Org/kilocode">
-          <img src="./public/providers/kilocode.png" alt="Kilo Code" width="48"/><br/>
-          <b>Kilo Code</b>
-        </a><br/>
-        <sub>⭐ 15.5K</sub>
-      </td>
-    </tr>
-  </table>
-
-<sub>📡 Todos los agentes se conectan vía <code>http://localhost:20128/v1</code> o <code>http://cloud.omniroute.online/v1</code> — una configuración, modelos y cuota ilimitados</sub>
-
----
-
-[![npm version](https://img.shields.io/npm/v/omniroute?color=cb3837&logo=npm)](https://www.npmjs.com/package/omniroute)
-[![Docker Hub](https://img.shields.io/docker/v/diegosouzapw/omniroute?label=Docker%20Hub&logo=docker&color=2496ED)](https://hub.docker.com/r/diegosouzapw/omniroute)
-[![License](https://img.shields.io/github/license/diegosouzapw/OmniRoute)](https://github.com/diegosouzapw/OmniRoute/blob/main/LICENSE)
-[![Website](https://img.shields.io/badge/Website-omniroute.online-blue?logo=google-chrome&logoColor=white)](https://omniroute.online)
-[![WhatsApp](https://img.shields.io/badge/WhatsApp-Community-25D366?logo=whatsapp&logoColor=white)](https://chat.whatsapp.com/JI7cDQ1GyaiDHhVBpLxf8b?mode=gi_t)
-
-[🌐 Website](https://omniroute.online) • [🚀 Inicio Rápido](#-inicio-rápido) • [💡 Características](#-características-principales) • [📖 Docs](#-documentación) • [💰 Precios](#-precios-resumidos)
-
-🌐 **Disponible en:** [English](README.md) | [Português](README.pt-BR.md) | [Español](README.es.md) | [Русский](README.ru.md) | [中文](README.zh-CN.md) | [Deutsch](README.de.md) | [Français](README.fr.md) | [Italiano](README.it.md)
-
-</div>
-
----
-
-## 🤔 ¿Por qué OmniRoute?
-
-**Deja de desperdiciar dinero y chocar con límites:**
-
-- <img src="https://img.shields.io/badge/✗-e74c3c?style=flat-square" height="16"/> La cuota de suscripción expira sin usar cada mes
-- <img src="https://img.shields.io/badge/✗-e74c3c?style=flat-square" height="16"/> Los límites de tasa te detienen en medio de la programación
-- <img src="https://img.shields.io/badge/✗-e74c3c?style=flat-square" height="16"/> APIs caras ($20-50/mes por proveedor)
-- <img src="https://img.shields.io/badge/✗-e74c3c?style=flat-square" height="16"/> Cambiar manualmente entre proveedores
-
-**OmniRoute resuelve esto:**
-
-- ✅ **Maximiza suscripciones** - Rastrea cuotas, usa cada bit antes del reset
-- ✅ **Fallback automático** - Suscripción → API Key → Barato → Gratuito, cero tiempo de inactividad
-- ✅ **Multi-cuenta** - Round-robin entre cuentas por proveedor
-- ✅ **Universal** - Funciona con Claude Code, Codex, Gemini CLI, Cursor, Cline, OpenClaw, cualquier herramienta CLI
-
----
-
-## 🔄 Cómo Funciona
-
-```
-┌─────────────┐
-│  Tu CLI     │  (Claude Code, Codex, Gemini CLI, OpenClaw, Cursor, Cline...)
-│   Tool      │
-└──────┬──────┘
-       │ http://localhost:20128/v1
-       ↓
-┌─────────────────────────────────────────┐
-│        OmniRoute (Enrutador Inteligente)  │
-│  • Traducción de formato (OpenAI ↔ Claude) │
-│  • Rastreo de cuota + Embeddings + Imágenes │
-│  • Renovación automática de tokens        │
-└──────┬──────────────────────────────────┘
-       │
-       ├─→ [Tier 1: SUSCRIPCIÓN] Claude Code, Codex, Gemini CLI
-       │   ↓ cuota agotada
-       ├─→ [Tier 2: API KEY] DeepSeek, Groq, xAI, Mistral, NVIDIA NIM, etc.
-       │   ↓ límite de presupuesto
-       ├─→ [Tier 3: BARATO] GLM ($0.6/1M), MiniMax ($0.2/1M)
-       │   ↓ límite de presupuesto
-       └─→ [Tier 4: GRATUITO] iFlow, Qwen, Kiro (ilimitado)
-
-Resultado: Nunca dejes de programar, costo mínimo
-```
-
----
-
-## ⚡ Inicio Rápido
-
-**1. Instala globalmente:**
-
-```bash
-npm install -g omniroute
-omniroute
-```
-
-🎉 El Dashboard se abre en `http://localhost:20128`
-
-| Comando                 | Descripción                                    |
-| ----------------------- | ---------------------------------------------- |
-| `omniroute`             | Iniciar servidor (puerto predeterminado 20128) |
-| `omniroute --port 3000` | Usar puerto personalizado                      |
-| `omniroute --no-open`   | No abrir navegador automáticamente             |
-| `omniroute --help`      | Mostrar ayuda                                  |
-
-**2. Conecta un proveedor GRATUITO:**
-
-Dashboard → Proveedores → Conectar **Claude Code** o **Antigravity** → Login OAuth → ¡Listo!
-
-**3. Usa en tu herramienta CLI:**
-
-```
-Claude Code/Codex/Gemini CLI/OpenClaw/Cursor/Cline Configuración:
-  Endpoint: http://localhost:20128/v1
-  API Key: [copiar del dashboard]
-  Model: if/kimi-k2-thinking
-```
-
-**¡Eso es todo!** Comienza a programar con modelos de IA GRATUITOS.
-
-**Alternativa — ejecutar desde código fuente:**
-
-```bash
-cp .env.example .env
-npm install
-PORT=20128 NEXT_PUBLIC_BASE_URL=http://localhost:20128 npm run dev
-```
-
----
-
-## 🐳 Docker
-
-OmniRoute está disponible como imagen Docker pública en [Docker Hub](https://hub.docker.com/r/diegosouzapw/omniroute).
-
-**Ejecución rápida:**
-
-```bash
-docker run -d \
-  --name omniroute \
-  --restart unless-stopped \
-  -p 20128:20128 \
-  -v omniroute-data:/app/data \
-  diegosouzapw/omniroute:latest
-```
-
-**Con archivo de entorno:**
-
-```bash
-# Copia y edita el .env primero
-cp .env.example .env
-
-docker run -d \
-  --name omniroute \
-  --restart unless-stopped \
-  --env-file .env \
-  -p 20128:20128 \
-  -v omniroute-data:/app/data \
-  diegosouzapw/omniroute:latest
-```
-
-**Usando Docker Compose:**
-
-```bash
-# Perfil base (sin herramientas CLI)
-docker compose --profile base up -d
-
-# Perfil CLI (Claude Code, Codex, OpenClaw integrados)
-docker compose --profile cli up -d
-```
-
-| Imagen                   | Tag      | Tamaño | Descripción            |
-| ------------------------ | -------- | ------ | ---------------------- |
-| `diegosouzapw/omniroute` | `latest` | ~250MB | Última versión estable |
-| `diegosouzapw/omniroute` | `1.0.6`  | ~250MB | Versión actual         |
-
----
-
-## 💰 Precios Resumidos
-
-| Tier               | Proveedor         | Costo                        | Reset de Cuota    | Mejor Para              |
-| ------------------ | ----------------- | ---------------------------- | ----------------- | ----------------------- |
-| **💳 SUSCRIPCIÓN** | Claude Code (Pro) | $20/mes                      | 5h + semanal      | Ya suscrito             |
-|                    | Codex (Plus/Pro)  | $20-200/mes                  | 5h + semanal      | Usuarios OpenAI         |
-|                    | Gemini CLI        | **GRATUITO**                 | 180K/mes + 1K/día | ¡Todos!                 |
-|                    | GitHub Copilot    | $10-19/mes                   | Mensual           | Usuarios GitHub         |
-| **🔑 API KEY**     | NVIDIA NIM        | **GRATUITO** (1000 créditos) | Único             | Pruebas gratuitas       |
-|                    | DeepSeek          | Por uso                      | Ninguno           | Mejor precio/calidad    |
-|                    | Groq              | Tier gratuito + pago         | Limitado          | Inferencia ultra-rápida |
-|                    | xAI (Grok)        | Por uso                      | Ninguno           | Modelos Grok            |
-|                    | Mistral           | Tier gratuito + pago         | Limitado          | IA Europea              |
-|                    | OpenRouter        | Por uso                      | Ninguno           | 100+ modelos            |
-| **💰 BARATO**      | GLM-4.7           | $0.6/1M                      | Diario 10h        | Respaldo económico      |
-|                    | MiniMax M2.1      | $0.2/1M                      | Rotativo 5h       | Opción más barata       |
-|                    | Kimi K2           | $9/mes fijo                  | 10M tokens/mes    | Costo predecible        |
-| **🆓 GRATUITO**    | iFlow             | $0                           | Ilimitado         | 8 modelos gratuitos     |
-|                    | Qwen              | $0                           | Ilimitado         | 3 modelos gratuitos     |
-|                    | Kiro              | $0                           | Ilimitado         | Claude gratuito         |
-
-**💡 Consejo Pro:** ¡Comienza con Gemini CLI (180K gratis/mes) + iFlow (ilimitado gratis) = $0 de costo!
-
----
-
-## 🎯 Casos de Uso
-
-### Caso 1: "Tengo suscripción Claude Pro"
-
-**Problema:** La cuota expira sin usar, límites de tasa durante programación intensa
-
-```
-Combo: "maximize-claude"
-  1. cc/claude-opus-4-6        (usar suscripción al máximo)
-  2. glm/glm-4.7               (respaldo barato cuando la cuota se agota)
-  3. if/kimi-k2-thinking       (fallback de emergencia gratuito)
-
-Costo mensual: $20 (suscripción) + ~$5 (respaldo) = $25 total
-vs. $20 + chocar con límites = frustración
-```
-
-### Caso 2: "Quiero costo cero"
-
-**Problema:** No puede pagar suscripciones, necesita IA confiable para programar
-
-```
-Combo: "free-forever"
-  1. gc/gemini-3-flash         (180K gratis/mes)
-  2. if/kimi-k2-thinking       (ilimitado gratis)
-  3. qw/qwen3-coder-plus       (ilimitado gratis)
-
-Costo mensual: $0
-Calidad: Modelos listos para producción
-```
-
-### Caso 3: "Necesito programar 24/7, sin interrupciones"
-
-**Problema:** Plazos ajustados, no puede permitirse tiempo de inactividad
-
-```
-Combo: "always-on"
-  1. cc/claude-opus-4-6        (mejor calidad)
-  2. cx/gpt-5.2-codex          (segunda suscripción)
-  3. glm/glm-4.7               (barato, reset diario)
-  4. minimax/MiniMax-M2.1      (más barato, reset 5h)
-  5. if/kimi-k2-thinking       (gratuito ilimitado)
-
-Resultado: 5 capas de fallback = cero tiempo de inactividad
-```
-
-### Caso 4: "Quiero IA GRATUITA en OpenClaw"
-
-**Problema:** Necesita asistente de IA en apps de mensajería, completamente gratuito
-
-```
-Combo: "openclaw-free"
-  1. if/glm-4.7                (ilimitado gratis)
-  2. if/minimax-m2.1           (ilimitado gratis)
-  3. if/kimi-k2-thinking       (ilimitado gratis)
-
-Costo mensual: $0
-Acceso vía: WhatsApp, Telegram, Slack, Discord, iMessage, Signal...
-```
-
----
-
-## 💡 Características Principales
-
-### 🧠 Enrutamiento e Inteligencia
-
-| Característica                         | Qué Hace                                                                        |
-| -------------------------------------- | ------------------------------------------------------------------------------- |
-| 🎯 **Fallback Inteligente 4 Tiers**    | Auto-enrutamiento: Suscripción → API Key → Barato → Gratuito                    |
-| 📊 **Rastreo de Cuota en Tiempo Real** | Conteo de tokens en vivo + countdown de reset por proveedor                     |
-| 🔄 **Traducción de Formato**           | OpenAI ↔ Claude ↔ Gemini ↔ Cursor ↔ Kiro transparente                           |
-| 👥 **Soporte Multi-Cuenta**            | Múltiples cuentas por proveedor con selección inteligente                       |
-| 🔄 **Renovación Automática de Token**  | Tokens OAuth se renuevan automáticamente con reintentos                         |
-| 🎨 **Combos Personalizados**           | 6 estrategias: fill-first, round-robin, p2c, random, least-used, cost-optimized |
-| 🧩 **Modelos Personalizados**          | Agrega cualquier ID de modelo a cualquier proveedor                             |
-| 🌐 **Enrutador Wildcard**              | Enruta patrones `provider/*` a cualquier proveedor dinámicamente                |
-| 🧠 **Presupuesto de Razonamiento**     | Modos passthrough, auto, custom y adaptativo para modelos de razonamiento       |
-| 💬 **Inyección de System Prompt**      | System prompt global aplicado en todas las solicitudes                          |
-| 📄 **API Responses**                   | Soporte completo de la API Responses de OpenAI (`/v1/responses`) para Codex     |
-
-### 🎵 APIs Multi-Modal
-
-| Característica                | Qué Hace                                               |
-| ----------------------------- | ------------------------------------------------------ |
-| 🖼️ **Generación de Imágenes** | `/v1/images/generations` — 4 proveedores, 9+ modelos   |
-| 📐 **Embeddings**             | `/v1/embeddings` — 6 proveedores, 9+ modelos           |
-| 🎤 **Transcripción de Audio** | `/v1/audio/transcriptions` — Compatible con Whisper    |
-| 🔊 **Texto a Voz**            | `/v1/audio/speech` — Síntesis de audio multi-proveedor |
-| 🛡️ **Moderaciones**           | `/v1/moderations` — Verificaciones de seguridad        |
-| 🔀 **Reranking**              | `/v1/rerank` — Reranking de relevancia de documentos   |
-
-### 🛡️ Resiliencia y Seguridad
-
-| Característica                     | Qué Hace                                                         |
-| ---------------------------------- | ---------------------------------------------------------------- |
-| 🔌 **Circuit Breaker**             | Auto-apertura/cierre por proveedor con umbrales configurables    |
-| 🛡️ **Anti-Thundering Herd**        | Mutex + semáforo rate-limit para proveedores con API key         |
-| 🧠 **Caché Semántico**             | Caché de dos niveles (firma + semántico) reduce costo y latencia |
-| ⚡ **Idempotencia de Solicitud**   | Ventana de dedup de 5s para solicitudes duplicadas               |
-| 🔒 **Spoofing de Fingerprint TLS** | Bypass de detección de bot vía TLS con wreq-js                   |
-| 🌐 **Filtrado de IP**              | Allowlist/blocklist para control de acceso a la API              |
-| 📊 **Rate Limits Editables**       | RPM, gap mínimo y concurrencia máxima configurables              |
-
-### 📊 Observabilidad y Analytics
-
-| Característica                 | Qué Hace                                                              |
-| ------------------------------ | --------------------------------------------------------------------- |
-| 📝 **Logs de Solicitud**       | Modo debug con logs completos de request/response                     |
-| 💾 **Logs SQLite**             | Logs de proxy persistentes sobreviven a reinicios                     |
-| 📊 **Dashboard de Analytics**  | Recharts: cards de estadísticas, gráfico de uso, tabla de proveedores |
-| 📈 **Rastreo de Progreso**     | Eventos de progreso SSE opt-in para streaming                         |
-| 🧪 **Evaluaciones de LLM**     | Pruebas con conjunto golden y 4 estrategias de match                  |
-| 🔍 **Telemetría de Solicitud** | Agregación de latencia p50/p95/p99 + rastreo X-Request-Id             |
-| 📋 **Logs + Cuotas**           | Páginas dedicadas para navegación de logs y rastreo de cuotas         |
-| 🏥 **Dashboard de Salud**      | Uptime, estados de circuit breaker, lockouts, stats de caché          |
-| 💰 **Rastreo de Costos**       | Gestión de presupuesto + configuración de precios por modelo          |
-
-### ☁️ Deploy y Sincronización
-
-| Característica                    | Qué Hace                                                                        |
-| --------------------------------- | ------------------------------------------------------------------------------- |
-| 💾 **Cloud Sync**                 | Sincroniza configuraciones entre dispositivos vía Cloudflare Workers            |
-| 🌐 **Deploy en Cualquier Lugar**  | Localhost, VPS, Docker, Cloudflare Workers                                      |
-| 🔑 **Gestión de API Keys**        | Genera, rota y define alcance de API keys por proveedor                         |
-| 🧙 **Asistente de Configuración** | Setup guiado en 4 pasos para nuevos usuarios                                    |
-| 🔧 **Dashboard CLI Tools**        | Configuración en un clic para Claude, Codex, Cline, OpenClaw, Kilo, Antigravity |
-| 🔄 **Backups de DB**              | Backup y restauración automáticos de todas las configuraciones                  |
-
-<details>
-<summary><b>📖 Detalles de Características</b></summary>
-
-### 🎯 Fallback Inteligente 4 Tiers
-
-Crea combos con fallback automático:
-
-```
-Combo: "my-coding-stack"
-  1. cc/claude-opus-4-6        (tu suscripción)
-  2. nvidia/llama-3.3-70b      (API NVIDIA gratuita)
-  3. glm/glm-4.7               (respaldo barato, $0.6/1M)
-  4. if/kimi-k2-thinking       (fallback gratuito)
-
-→ Cambia automáticamente cuando la cuota se agota o ocurren errores
-```
-
-### 📊 Rastreo de Cuota en Tiempo Real
-
-- Consumo de tokens por proveedor
-- Countdown de reset (5 horas, diario, semanal)
-- Estimación de costo para tiers pagos
-- Reportes de gastos mensuales
-
-### 🔄 Traducción de Formato
-
-Traducción transparente entre formatos:
-
-- **OpenAI** ↔ **Claude** ↔ **Gemini** ↔ **OpenAI Responses**
-- Tu herramienta CLI envía formato OpenAI → OmniRoute traduce → El proveedor recibe formato nativo
-- Funciona con cualquier herramienta que soporte endpoints OpenAI personalizados
-
-### 👥 Soporte Multi-Cuenta
-
-- Agrega múltiples cuentas por proveedor
-- Round-robin automático o enrutamiento por prioridad
-- Fallback a la siguiente cuenta cuando una alcanza la cuota
-
-### 🔄 Renovación Automática de Token
-
-- Los tokens OAuth se renuevan automáticamente antes de expirar
-- Sin necesidad de re-autenticación manual
-- Experiencia transparente en todos los proveedores
-
-### 🎨 Combos Personalizados
-
-- Crea combinaciones ilimitadas de modelos
-- 6 estrategias: fill-first, round-robin, power-of-two-choices, random, least-used, cost-optimized
-- Comparte combos entre dispositivos con Cloud Sync
-
-### 🏥 Dashboard de Salud
-
-- Estado del sistema (uptime, versión, uso de memoria)
-- Estados de circuit breaker por proveedor (Closed/Open/Half-Open)
-- Estado de rate limit y lockouts activos
-- Estadísticas de caché de firma
-- Telemetría de latencia (p50/p95/p99) + caché de prompt
-- Reset de salud con un clic
-
-### 🔧 Playground del Traductor
-
-- Debug, prueba y visualiza traducciones de formato de API
-- Envía solicitudes y ve cómo OmniRoute traduce entre formatos de proveedores
-- Invaluable para troubleshooting de problemas de integración
-
-### 💾 Cloud Sync
-
-- Sincroniza proveedores, combos y configuraciones entre dispositivos
-- Sincronización automática en segundo plano
-- Almacenamiento cifrado seguro
-
-</details>
-
----
-
-## 📖 Guía de Configuración
-
-<details>
-<summary><b>💳 Proveedores por Suscripción</b></summary>
-
-### Claude Code (Pro/Max)
-
-```bash
-Dashboard → Proveedores → Conectar Claude Code
-→ Login OAuth → Renovación automática de token
-→ Rastreo de cuota 5h + semanal
-
-Modelos:
-  cc/claude-opus-4-6
-  cc/claude-sonnet-4-5-20250929
-  cc/claude-haiku-4-5-20251001
-```
-
-**Consejo Pro:** Usa Opus para tareas complejas, Sonnet para velocidad. ¡OmniRoute rastrea cuota por modelo!
-
-### OpenAI Codex (Plus/Pro)
-
-```bash
-Dashboard → Proveedores → Conectar Codex
-→ Login OAuth (puerto 1455)
-→ Reset 5h + semanal
-
-Modelos:
-  cx/gpt-5.2-codex
-  cx/gpt-5.1-codex-max
-```
-
-### Gemini CLI (¡GRATUITO 180K/mes!)
-
-```bash
-Dashboard → Proveedores → Conectar Gemini CLI
-→ Google OAuth
-→ 180K completions/mes + 1K/día
-
-Modelos:
-  gc/gemini-3-flash-preview
-  gc/gemini-2.5-pro
-```
-
-**Mejor Valor:** ¡Tier gratuito enorme! Úsalo antes de los tiers pagos.
-
-### GitHub Copilot
-
-```bash
-Dashboard → Proveedores → Conectar GitHub
-→ OAuth vía GitHub
-→ Reset mensual (1ro del mes)
-
-Modelos:
-  gh/gpt-5
-  gh/claude-4.5-sonnet
-  gh/gemini-3-pro
-```
-
-</details>
-
-<details>
-<summary><b>🔑 Proveedores por API Key</b></summary>
-
-### NVIDIA NIM (¡GRATUITO 1000 créditos!)
-
-1. Regístrate: [build.nvidia.com](https://build.nvidia.com)
-2. Obtén API key gratuita (1000 créditos de inferencia incluidos)
-3. Dashboard → Agregar Proveedor → NVIDIA NIM:
-   - API Key: `nvapi-your-key`
-
-**Modelos:** `nvidia/llama-3.3-70b-instruct`, `nvidia/mistral-7b-instruct`, y 50+ más
-
-**Consejo Pro:** ¡API compatible con OpenAI — funciona perfectamente con la traducción de formato de OmniRoute!
-
-### DeepSeek
-
-1. Regístrate: [platform.deepseek.com](https://platform.deepseek.com)
-2. Obtén API key
-3. Dashboard → Agregar Proveedor → DeepSeek
-
-**Modelos:** `deepseek/deepseek-chat`, `deepseek/deepseek-coder`
-
-### Groq (¡Tier Gratuito Disponible!)
-
-1. Regístrate: [console.groq.com](https://console.groq.com)
-2. Obtén API key (tier gratuito incluido)
-3. Dashboard → Agregar Proveedor → Groq
-
-**Modelos:** `groq/llama-3.3-70b`, `groq/mixtral-8x7b`
-
-**Consejo Pro:** ¡Inferencia ultra-rápida — mejor para programación en tiempo real!
-
-### OpenRouter (100+ Modelos)
-
-1. Regístrate: [openrouter.ai](https://openrouter.ai)
-2. Obtén API key
-3. Dashboard → Agregar Proveedor → OpenRouter
-
-**Modelos:** Accede a 100+ modelos de todos los principales proveedores a través de una única API key.
-
-</details>
-
-<details>
-<summary><b>💰 Proveedores Baratos (Respaldo)</b></summary>
-
-### GLM-4.7 (Reset diario, $0.6/1M)
-
-1. Regístrate: [Zhipu AI](https://open.bigmodel.cn/)
-2. Obtén API key del Plan Coding
-3. Dashboard → Agregar API Key:
-   - Proveedor: `glm`
-   - API Key: `your-key`
-
-**Usa:** `glm/glm-4.7`
-
-**Consejo Pro:** ¡El Plan Coding ofrece 3× cuota a 1/7 del costo! Reset diario 10:00 AM.
-
-### MiniMax M2.1 (Reset 5h, $0.20/1M)
-
-1. Regístrate: [MiniMax](https://www.minimax.io/)
-2. Obtén API key
-3. Dashboard → Agregar API Key
-
-**Usa:** `minimax/MiniMax-M2.1`
-
-**Consejo Pro:** ¡Opción más barata para contexto largo (1M tokens)!
-
-### Kimi K2 ($9/mes fijo)
-
-1. Suscríbete: [Moonshot AI](https://platform.moonshot.ai/)
-2. Obtén API key
-3. Dashboard → Agregar API Key
-
-**Usa:** `kimi/kimi-latest`
-
-**Consejo Pro:** ¡$9/mes fijo por 10M tokens = $0.90/1M de costo efectivo!
-
-</details>
-
-<details>
-<summary><b>🆓 Proveedores GRATUITOS (Respaldo de Emergencia)</b></summary>
-
-### iFlow (8 modelos GRATUITOS)
-
-```bash
-Dashboard → Conectar iFlow
-→ Login OAuth iFlow
-→ Uso ilimitado
-
-Modelos:
-  if/kimi-k2-thinking
-  if/qwen3-coder-plus
-  if/glm-4.7
-  if/minimax-m2
-  if/deepseek-r1
-```
-
-### Qwen (3 modelos GRATUITOS)
-
-```bash
-Dashboard → Conectar Qwen
-→ Autorización por código de dispositivo
-→ Uso ilimitado
-
-Modelos:
-  qw/qwen3-coder-plus
-  qw/qwen3-coder-flash
-```
-
-### Kiro (Claude GRATUITO)
-
-```bash
-Dashboard → Conectar Kiro
-→ AWS Builder ID o Google/GitHub
-→ Uso ilimitado
-
-Modelos:
-  kr/claude-sonnet-4.5
-  kr/claude-haiku-4.5
-```
-
-</details>
-
-<details>
-<summary><b>🎨 Crear Combos</b></summary>
-
-### Ejemplo 1: Maximizar Suscripción → Respaldo Barato
-
-```
-Dashboard → Combos → Crear Nuevo
-
-Nombre: premium-coding
-Modelos:
-  1. cc/claude-opus-4-6 (Suscripción primaria)
-  2. glm/glm-4.7 (Respaldo barato, $0.6/1M)
-  3. minimax/MiniMax-M2.1 (Fallback más barato, $0.20/1M)
-
-Usa en CLI: premium-coding
-```
-
-### Ejemplo 2: Solo Gratuito (Costo Cero)
-
-```
-Nombre: free-combo
-Modelos:
-  1. gc/gemini-3-flash-preview (180K gratis/mes)
-  2. if/kimi-k2-thinking (ilimitado)
-  3. qw/qwen3-coder-plus (ilimitado)
-
-Costo: ¡$0 para siempre!
-```
-
-</details>
-
-<details>
-<summary><b>🔧 Integración CLI</b></summary>
-
-### Cursor IDE
-
-```
-Configuración → Modelos → Avanzado:
-  OpenAI API Base URL: http://localhost:20128/v1
-  OpenAI API Key: [del dashboard OmniRoute]
-  Model: cc/claude-opus-4-6
-```
-
-### Claude Code
-
-Usa la página **CLI Tools** en el dashboard para configuración en un clic, o edita `~/.claude/settings.json` manualmente.
-
-### Codex CLI
-
-```bash
-export OPENAI_BASE_URL="http://localhost:20128"
-export OPENAI_API_KEY="your-omniroute-api-key"
-
-codex "your prompt"
-```
-
-### OpenClaw
-
-**Opción 1 — Dashboard (recomendado):**
-
-```
-Dashboard → CLI Tools → OpenClaw → Seleccionar Modelo → Aplicar
-```
-
-**Opción 2 — Manual:** Edita `~/.openclaw/openclaw.json`:
-
-```json
-{
-  "models": {
-    "providers": {
-      "omniroute": {
-        "baseUrl": "http://127.0.0.1:20128/v1",
-        "apiKey": "sk_omniroute",
-        "api": "openai-completions"
-      }
-    }
-  }
-}
-```
-
-> **Nota:** OpenClaw solo funciona con OmniRoute local. Usa `127.0.0.1` en lugar de `localhost` para evitar problemas de resolución IPv6.
-
-### Cline / Continue / RooCode
-
-```
-Configuración → Configuración de API:
-  Proveedor: OpenAI Compatible
-  Base URL: http://localhost:20128/v1
-  API Key: [del dashboard OmniRoute]
-  Model: if/kimi-k2-thinking
-```
-
-</details>
-
----
-
-## 📊 Modelos Disponibles
-
-<details>
-<summary><b>Ver todos los modelos disponibles</b></summary>
-
-**Claude Code (`cc/`)** - Pro/Max:
-
-- `cc/claude-opus-4-6`
-- `cc/claude-sonnet-4-5-20250929`
-- `cc/claude-haiku-4-5-20251001`
-
-**Codex (`cx/`)** - Plus/Pro:
-
-- `cx/gpt-5.2-codex`
-- `cx/gpt-5.1-codex-max`
-
-**Gemini CLI (`gc/`)** - GRATUITO:
-
-- `gc/gemini-3-flash-preview`
-- `gc/gemini-2.5-pro`
-
-**GitHub Copilot (`gh/`)**:
-
-- `gh/gpt-5`
-- `gh/claude-4.5-sonnet`
-
-**NVIDIA NIM (`nvidia/`)** - Créditos GRATUITOS:
-
-- `nvidia/llama-3.3-70b-instruct`
-- `nvidia/mistral-7b-instruct`
-- 50+ más modelos en [build.nvidia.com](https://build.nvidia.com)
-
-**GLM (`glm/`)** - $0.6/1M:
-
-- `glm/glm-4.7`
-
-**MiniMax (`minimax/`)** - $0.2/1M:
-
-- `minimax/MiniMax-M2.1`
-
-**iFlow (`if/`)** - GRATUITO:
-
-- `if/kimi-k2-thinking`
-- `if/qwen3-coder-plus`
-- `if/deepseek-r1`
-- `if/glm-4.7`
-- `if/minimax-m2`
-
-**Qwen (`qw/`)** - GRATUITO:
-
-- `qw/qwen3-coder-plus`
-- `qw/qwen3-coder-flash`
-
-**Kiro (`kr/`)** - GRATUITO:
-
-- `kr/claude-sonnet-4.5`
-- `kr/claude-haiku-4.5`
-
-**OpenRouter (`or/`)** - 100+ modelos:
-
-- `or/anthropic/claude-4-sonnet`
-- `or/google/gemini-2.5-pro`
-- Cualquier modelo de [openrouter.ai/models](https://openrouter.ai/models)
-
-</details>
-
----
-
-## 🧪 Evaluaciones (Evals)
-
-OmniRoute incluye un framework de evaluación integrado para probar la calidad de respuestas de LLM contra un conjunto golden. Accede vía **Analytics → Evals** en el dashboard.
-
-### Conjunto Golden Integrado
-
-El "OmniRoute Golden Set" precargado contiene 10 casos de prueba que cubren:
-
-- Saludos, matemáticas, geografía, generación de código
-- Conformidad de formato JSON, traducción, markdown
-- Rechazo de seguridad (contenido dañino), conteo, lógica booleana
-
-### Estrategias de Evaluación
-
-| Estrategia | Descripción                                          | Ejemplo                          |
-| ---------- | ---------------------------------------------------- | -------------------------------- |
-| `exact`    | La salida debe coincidir exactamente                 | `"4"`                            |
-| `contains` | La salida debe contener subcadena (case-insensitive) | `"Paris"`                        |
-| `regex`    | La salida debe coincidir con el patrón regex         | `"1.*2.*3"`                      |
-| `custom`   | Función JS personalizada retorna true/false          | `(output) => output.length > 10` |
-
----
-
-## 🐛 Solución de Problemas
-
-<details>
-<summary><b>Haz clic para expandir la guía de solución de problemas</b></summary>
-
-**"Language model did not provide messages"**
-
-- Cuota del proveedor agotada → Verifica el rastreador de cuota en el dashboard
-- Solución: Usa combo con fallback o cambia a tier más barato
-
-**Rate limiting**
-
-- Cuota de suscripción agotada → Fallback a GLM/MiniMax
-- Agrega combo: `cc/claude-opus-4-6 → glm/glm-4.7 → if/kimi-k2-thinking`
-
-**Token OAuth expirado**
-
-- Renovado automáticamente por OmniRoute
-- Si persiste: Dashboard → Proveedor → Reconectar
-
-**Costos altos**
-
-- Verifica estadísticas de uso en Dashboard → Costos
-- Cambia modelo primario a GLM/MiniMax
-- Usa tier gratuito (Gemini CLI, iFlow) para tareas no críticas
-
-**Dashboard se abre en el puerto equivocado**
-
-- Establece `PORT=20128` y `NEXT_PUBLIC_BASE_URL=http://localhost:20128`
-
-**Errores de cloud sync**
-
-- Verifica que `BASE_URL` apunte a tu instancia en ejecución
-- Verifica que `CLOUD_URL` apunte a tu endpoint cloud esperado
-- Mantén los valores `NEXT_PUBLIC_*` alineados con los valores del servidor
-
-**Primer login no funciona**
-
-- Verifica `INITIAL_PASSWORD` en `.env`
-- Si no está definido, la contraseña predeterminada es `123456`
-
-**Sin logs de solicitud**
-
-- Establece `ENABLE_REQUEST_LOGS=true` en `.env`
-
-**Prueba de conexión muestra "Invalid" para proveedores compatibles con OpenAI**
-
-- Muchos proveedores no exponen el endpoint `/models`
-- OmniRoute v1.0.6+ incluye validación vía chat completions como fallback
-- Asegúrate de que la URL base incluya el sufijo `/v1`
-
-</details>
-
----
-
-## 🛠️ Stack Tecnológico
-
-- **Runtime**: Node.js 20+
-- **Lenguaje**: TypeScript 5.9 — **100% TypeScript** en `src/` y `open-sse/` (v1.0.6)
-- **Framework**: Next.js 16 + React 19 + Tailwind CSS 4
-- **Base de Datos**: LowDB (JSON) + SQLite (estado del dominio + logs de proxy)
-- **Streaming**: Server-Sent Events (SSE)
-- **Auth**: OAuth 2.0 (PKCE) + JWT + API Keys
-- **Testing**: Node.js test runner (368+ tests unitarios)
-- **CI/CD**: GitHub Actions (publicación automática npm + Docker Hub en release)
-- **Website**: [omniroute.online](https://omniroute.online)
-- **Paquete**: [npmjs.com/package/omniroute](https://www.npmjs.com/package/omniroute)
-- **Docker**: [hub.docker.com/r/diegosouzapw/omniroute](https://hub.docker.com/r/diegosouzapw/omniroute)
-- **Resiliencia**: Circuit breaker, backoff exponencial, anti-thundering herd, spoofing TLS
-
----
-
-## 📖 Documentación
-
-| Documento                                        | Descripción                                        |
-| ------------------------------------------------ | -------------------------------------------------- |
-| [Guía del Usuario](docs/USER_GUIDE.md)           | Proveedores, combos, integración CLI, deploy       |
-| [Referencia de API](docs/API_REFERENCE.md)       | Todos los endpoints con ejemplos                   |
-| [Solución de Problemas](docs/TROUBLESHOOTING.md) | Problemas comunes y soluciones                     |
-| [Arquitectura](docs/ARCHITECTURE.md)             | Arquitectura del sistema e internos                |
-| [Contribuir](CONTRIBUTING.md)                    | Setup de desarrollo y directrices                  |
-| [Spec OpenAPI](docs/openapi.yaml)                | Especificación OpenAPI 3.0                         |
-| [Política de Seguridad](SECURITY.md)             | Reportar vulnerabilidades y prácticas de seguridad |
-
----
-
-## 📧 Soporte
-
-> 💬 **¡Únete a la comunidad!** [Grupo WhatsApp](https://chat.whatsapp.com/JI7cDQ1GyaiDHhVBpLxf8b?mode=gi_t) — Obtén ayuda, comparte consejos y mantente al día.
-
-- **Website**: [omniroute.online](https://omniroute.online)
-- **GitHub**: [github.com/diegosouzapw/OmniRoute](https://github.com/diegosouzapw/OmniRoute)
-- **Issues**: [github.com/diegosouzapw/OmniRoute/issues](https://github.com/diegosouzapw/OmniRoute/issues)
-- **WhatsApp**: [Grupo de la Comunidad](https://chat.whatsapp.com/JI7cDQ1GyaiDHhVBpLxf8b?mode=gi_t)
-- **Proyecto Original**: [9router por decolua](https://github.com/decolua/9router)
-
----
-
-## 👥 Contribuidores
-
-[![Contributors](https://contrib.rocks/image?repo=diegosouzapw/OmniRoute&max=100&columns=20&anon=1)](https://github.com/diegosouzapw/OmniRoute/graphs/contributors)
-
-### Cómo Contribuir
-
-1. Haz fork del repositorio
-2. Crea tu rama de funcionalidad (`git checkout -b feature/amazing-feature`)
-3. Haz commit de tus cambios (`git commit -m 'Add amazing feature'`)
-4. Haz push a la rama (`git push origin feature/amazing-feature`)
-5. Abre un Pull Request
-
-Consulta [CONTRIBUTING.md](CONTRIBUTING.md) para directrices detalladas.
-
-### Lanzar una Nueva Versión
-
-```bash
-# Crea un release — la publicación en npm ocurre automáticamente
-gh release create v1.0.6 --title "v1.0.6" --generate-notes
-```
-
----
-
-## 📊 Historial de Stars
-
-<a href="https://star-history.com/#diegosouzapw/OmniRoute&Date">
- <picture>
-   <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/svg?repos=diegosouzapw/OmniRoute&type=Date&theme=dark" />
-   <source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/svg?repos=diegosouzapw/OmniRoute&type=Date" />
-   <img alt="Star History Chart" src="https://api.star-history.com/svg?repos=diegosouzapw/OmniRoute&type=Date" />
- </picture>
-</a>
-
----
-
-## 🙏 Agradecimientos
-
-Agradecimiento especial a **[9router](https://github.com/decolua/9router)** por **[decolua](https://github.com/decolua)** — el proyecto original que inspiró este fork. OmniRoute se construye sobre esa increíble base con características adicionales, APIs multi-modal y una reescritura completa en TypeScript.
-
-Agradecimiento especial a **[CLIProxyAPI](https://github.com/router-for-me/CLIProxyAPI)** — la implementación original en Go que inspiró esta adaptación a JavaScript.
-
----
-
-## 📄 Licencia
-
-Licencia MIT - consulta [LICENSE](LICENSE) para detalles.
-
----
-
-<div align="center">
-  <sub>Hecho con ❤️ para desarrolladores que programan 24/7</sub>
-  <br/>
-  <sub><a href="https://omniroute.online">omniroute.online</a></sub>
-</div>

README.fr.md

@@ -1,999 +0,0 @@
-<div align="center">
-  <img src="./docs/screenshots/MainOmniRoute.png" alt="OmniRoute Dashboard" width="800"/>
-  
-  # 🚀 OmniRoute — La Passerelle IA Gratuite
-
-### N'arrêtez jamais de coder. Routage intelligent vers des **modèles IA GRATUITS et économiques** avec fallback automatique.
-
-_Votre proxy API universel — un endpoint, 36+ fournisseurs, zéro temps d'arrêt._
-
-**Chat Completions • Embeddings • Génération d'images • Audio • Reranking • 100% TypeScript**
-
----
-
-### 🤖 Fournisseur IA gratuit pour vos agents de programmation préférés
-
-_Connectez n'importe quel IDE ou outil CLI alimenté par l'IA via OmniRoute — passerelle API gratuite pour un codage illimité._
-
-  <table>
-    <tr>
-      <td align="center" width="110">
-        <a href="https://github.com/cline/cline">
-          <img src="./public/providers/openclaw.png" alt="OpenClaw" width="48"/><br/>
-          <b>OpenClaw</b>
-        </a><br/>
-        <sub>⭐ 205K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/HKUDS/nanobot">
-          <img src="./public/providers/nanobot.png" alt="NanoBot" width="48"/><br/>
-          <b>NanoBot</b>
-        </a><br/>
-        <sub>⭐ 20.9K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/sipeed/picoclaw">
-          <img src="./public/providers/picoclaw.jpg" alt="PicoClaw" width="48"/><br/>
-          <b>PicoClaw</b>
-        </a><br/>
-        <sub>⭐ 14.6K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/zeroclaw-labs/zeroclaw">
-          <img src="./public/providers/zeroclaw.png" alt="ZeroClaw" width="48"/><br/>
-          <b>ZeroClaw</b>
-        </a><br/>
-        <sub>⭐ 9.9K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/nearai/ironclaw">
-          <img src="./public/providers/ironclaw.png" alt="IronClaw" width="48"/><br/>
-          <b>IronClaw</b>
-        </a><br/>
-        <sub>⭐ 2.1K</sub>
-      </td>
-    </tr>
-    <tr>
-      <td align="center" width="110">
-        <a href="https://github.com/anomalyco/opencode">
-          <img src="./public/providers/opencode.svg" alt="OpenCode" width="48"/><br/>
-          <b>OpenCode</b>
-        </a><br/>
-        <sub>⭐ 106K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/openai/codex">
-          <img src="./public/providers/codex.png" alt="Codex CLI" width="48"/><br/>
-          <b>Codex CLI</b>
-        </a><br/>
-        <sub>⭐ 60.8K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/anthropics/claude-code">
-          <img src="./public/providers/claude.png" alt="Claude Code" width="48"/><br/>
-          <b>Claude Code</b>
-        </a><br/>
-        <sub>⭐ 67.3K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/google-gemini/gemini-cli">
-          <img src="./public/providers/gemini-cli.png" alt="Gemini CLI" width="48"/><br/>
-          <b>Gemini CLI</b>
-        </a><br/>
-        <sub>⭐ 94.7K</sub>
-      </td>
-      <td align="center" width="110">
-        <a href="https://github.com/Kilo-Org/kilocode">
-          <img src="./public/providers/kilocode.png" alt="Kilo Code" width="48"/><br/>
-          <b>Kilo Code</b>
-        </a><br/>
-        <sub>⭐ 15.5K</sub>
-      </td>
-    </tr>
-  </table>
-
-<sub>📡 Tous les agents se connectent via <code>http://localhost:20128/v1</code> ou <code>http://cloud.omniroute.online/v1</code> — une configuration, modèles et quota illimités</sub>
-
----
-
-[![npm version](https://img.shields.io/npm/v/omniroute?color=cb3837&logo=npm)](https://www.npmjs.com/package/omniroute)
-[![Docker Hub](https://img.shields.io/docker/v/diegosouzapw/omniroute?label=Docker%20Hub&logo=docker&color=2496ED)](https://hub.docker.com/r/diegosouzapw/omniroute)
-[![License](https://img.shields.io/github/license/diegosouzapw/OmniRoute)](https://github.com/diegosouzapw/OmniRoute/blob/main/LICENSE)
-[![Website](https://img.shields.io/badge/Website-omniroute.online-blue?logo=google-chrome&logoColor=white)](https://omniroute.online)
-[![WhatsApp](https://img.shields.io/badge/WhatsApp-Community-25D366?logo=whatsapp&logoColor=white)](https://chat.whatsapp.com/JI7cDQ1GyaiDHhVBpLxf8b?mode=gi_t)
-
-[🌐 Site web](https://omniroute.online) • [🚀 Démarrage rapide](#-démarrage-rapide) • [💡 Fonctionnalités](#-fonctionnalités-principales) • [📖 Docs](#-documentation) • [💰 Tarifs](#-aperçu-des-tarifs)
-
-🌐 **Disponible en :** [English](README.md) | [Português](README.pt-BR.md) | [Español](README.es.md) | [Русский](README.ru.md) | [中文](README.zh-CN.md) | [Deutsch](README.de.md) | [Français](README.fr.md) | [Italiano](README.it.md)
-
-</div>
-
----
-
-## 🤔 Pourquoi OmniRoute ?
-
-**Arrêtez de gaspiller de l'argent et de vous heurter aux limites :**
-
-- <img src="https://img.shields.io/badge/✗-e74c3c?style=flat-square" height="16"/> Le quota d'abonnement expire inutilisé chaque mois
-- <img src="https://img.shields.io/badge/✗-e74c3c?style=flat-square" height="16"/> Les limites de débit vous arrêtent en plein codage
-- <img src="https://img.shields.io/badge/✗-e74c3c?style=flat-square" height="16"/> APIs coûteuses (20-50 $/mois par fournisseur)
-- <img src="https://img.shields.io/badge/✗-e74c3c?style=flat-square" height="16"/> Changement manuel entre fournisseurs
-
-**OmniRoute résout ces problèmes :**
-
-- ✅ **Maximisez les abonnements** — Suivez les quotas, utilisez chaque bit avant la réinitialisation
-- ✅ **Fallback automatique** — Abonnement → Clé API → Économique → Gratuit, zéro temps d'arrêt
-- ✅ **Multi-comptes** — Round-robin entre les comptes par fournisseur
-- ✅ **Universel** — Fonctionne avec Claude Code, Codex, Gemini CLI, Cursor, Cline, OpenClaw, tout outil CLI
-
----
-
-## 🔄 Comment ça fonctionne
-
-```
-┌─────────────┐
-│  Votre CLI  │  (Claude Code, Codex, Gemini CLI, OpenClaw, Cursor, Cline...)
-│   Tool      │
-└──────┬──────┘
-       │ http://localhost:20128/v1
-       ↓
-┌─────────────────────────────────────────┐
-│        OmniRoute (Routeur intelligent)   │
-│  • Traduction de format (OpenAI ↔ Claude) │
-│  • Suivi des quotas + Embeddings + Images │
-│  • Renouvellement automatique des tokens   │
-└──────┬──────────────────────────────────┘
-       │
-       ├─→ [Tier 1: ABONNEMENT] Claude Code, Codex, Gemini CLI
-       │   ↓ quota épuisé
-       ├─→ [Tier 2: CLÉ API] DeepSeek, Groq, xAI, Mistral, NVIDIA NIM, etc.
-       │   ↓ limite de budget
-       ├─→ [Tier 3: ÉCONOMIQUE] GLM ($0.6/1M), MiniMax ($0.2/1M)
-       │   ↓ limite de budget
-       └─→ [Tier 4: GRATUIT] iFlow, Qwen, Kiro (illimité)
-
-Résultat : Ne jamais arrêter de coder, coût minimal
-```
-
----
-
-## ⚡ Démarrage rapide
-
-**1. Installer globalement :**
-
-```bash
-npm install -g omniroute
-omniroute
-```
-
-🎉 Le tableau de bord s'ouvre sur `http://localhost:20128`
-
-| Commande                | Description                                 |
-| ----------------------- | ------------------------------------------- |
-| `omniroute`             | Démarrer le serveur (port par défaut 20128) |
-| `omniroute --port 3000` | Utiliser un port personnalisé               |
-| `omniroute --no-open`   | Ne pas ouvrir le navigateur automatiquement |
-| `omniroute --help`      | Afficher l'aide                             |
-
-**2. Connecter un fournisseur GRATUIT :**
-
-Tableau de bord → Fournisseurs → Connecter **Claude Code** ou **Antigravity** → Connexion OAuth → Terminé !
-
-**3. Utiliser dans votre outil CLI :**
-
-```
-Claude Code/Codex/Gemini CLI/OpenClaw/Cursor/Cline Paramètres :
-  Endpoint : http://localhost:20128/v1
-  API Key : [copier depuis le tableau de bord]
-  Model : if/kimi-k2-thinking
-```
-
-**C'est tout !** Commencez à coder avec des modèles IA GRATUITS.
-
-**Alternative — exécuter depuis le code source :**
-
-```bash
-cp .env.example .env
-npm install
-PORT=20128 NEXT_PUBLIC_BASE_URL=http://localhost:20128 npm run dev
-```
-
----
-
-## 🐳 Docker
-
-OmniRoute est disponible en tant qu'image Docker publique sur [Docker Hub](https://hub.docker.com/r/diegosouzapw/omniroute).
-
-**Démarrage rapide :**
-
-```bash
-docker run -d \
-  --name omniroute \
-  --restart unless-stopped \
-  -p 20128:20128 \
-  -v omniroute-data:/app/data \
-  diegosouzapw/omniroute:latest
-```
-
-**Avec fichier d'environnement :**
-
-```bash
-# Copier et modifier le .env d'abord
-cp .env.example .env
-
-docker run -d \
-  --name omniroute \
-  --restart unless-stopped \
-  --env-file .env \
-  -p 20128:20128 \
-  -v omniroute-data:/app/data \
-  diegosouzapw/omniroute:latest
-```
-
-**Avec Docker Compose :**
-
-```bash
-# Profil de base (sans outils CLI)
-docker compose --profile base up -d
-
-# Profil CLI (Claude Code, Codex, OpenClaw intégrés)
-docker compose --profile cli up -d
-```
-
-| Image                    | Tag      | Taille | Description             |
-| ------------------------ | -------- | ------ | ----------------------- |
-| `diegosouzapw/omniroute` | `latest` | ~250MB | Dernière version stable |
-| `diegosouzapw/omniroute` | `1.0.6`  | ~250MB | Version actuelle        |
-
----
-
-## 💰 Aperçu des tarifs
-
-| Tier              | Fournisseur       | Coût                       | Réinitialisation    | Idéal pour                    |
-| ----------------- | ----------------- | -------------------------- | ------------------- | ----------------------------- |
-| **💳 ABONNEMENT** | Claude Code (Pro) | 20 $/mois                  | 5h + hebdomadaire   | Déjà abonné                   |
-|                   | Codex (Plus/Pro)  | 20-200 $/mois              | 5h + hebdomadaire   | Utilisateurs OpenAI           |
-|                   | Gemini CLI        | **GRATUIT**                | 180K/mois + 1K/jour | Tout le monde !               |
-|                   | GitHub Copilot    | 10-19 $/mois               | Mensuel             | Utilisateurs GitHub           |
-| **🔑 CLÉ API**    | NVIDIA NIM        | **GRATUIT** (1000 crédits) | Unique              | Tests gratuits                |
-|                   | DeepSeek          | À l'usage                  | Aucune              | Meilleur rapport qualité-prix |
-|                   | Groq              | Niveau gratuit + payant    | Limité              | Inférence ultra-rapide        |
-|                   | xAI (Grok)        | À l'usage                  | Aucune              | Modèles Grok                  |
-|                   | Mistral           | Niveau gratuit + payant    | Limité              | IA européenne                 |
-|                   | OpenRouter        | À l'usage                  | Aucune              | 100+ modèles                  |
-| **💰 ÉCONOMIQUE** | GLM-4.7           | 0,6 $/1M                   | Quotidien 10h       | Backup économique             |
-|                   | MiniMax M2.1      | 0,2 $/1M                   | Rotatif 5h          | Option la moins chère         |
-|                   | Kimi K2           | 9 $/mois fixe              | 10M tokens/mois     | Coût prévisible               |
-| **🆓 GRATUIT**    | iFlow             | 0 $                        | Illimité            | 8 modèles gratuits            |
-|                   | Qwen              | 0 $                        | Illimité            | 3 modèles gratuits            |
-|                   | Kiro              | 0 $                        | Illimité            | Claude gratuit                |
-
-**💡 Conseil Pro :** Commencez avec Gemini CLI (180K gratuits/mois) + iFlow (illimité gratuit) = 0 $ de coût !
-
----
-
-## 🎯 Cas d'utilisation
-
-### Cas 1 : « J'ai un abonnement Claude Pro »
-
-**Problème :** Le quota expire inutilisé, limites de débit pendant le codage intensif
-
-```
-Combo : "maximize-claude"
-  1. cc/claude-opus-4-6        (utiliser l'abonnement au maximum)
-  2. glm/glm-4.7               (backup économique quand le quota est épuisé)
-  3. if/kimi-k2-thinking       (fallback d'urgence gratuit)
-
-Coût mensuel : 20 $ (abonnement) + ~5 $ (backup) = 25 $ au total
-vs. 20 $ + atteindre les limites = frustration
-```
-
-### Cas 2 : « Je veux zéro coût »
-
-**Problème :** Impossible de payer des abonnements, besoin d'IA fiable pour coder
-
-```
-Combo : "free-forever"
-  1. gc/gemini-3-flash         (180K gratuits/mois)
-  2. if/kimi-k2-thinking       (illimité gratuit)
-  3. qw/qwen3-coder-plus       (illimité gratuit)
-
-Coût mensuel : 0 $
-Qualité : Modèles prêts pour la production
-```
-
-### Cas 3 : « Je dois coder 24/7, sans interruption »
-
-**Problème :** Délais serrés, ne peut pas se permettre de temps d'arrêt
-
-```
-Combo : "always-on"
-  1. cc/claude-opus-4-6        (meilleure qualité)
-  2. cx/gpt-5.2-codex          (deuxième abonnement)
-  3. glm/glm-4.7               (économique, reset quotidien)
-  4. minimax/MiniMax-M2.1      (le moins cher, reset 5h)
-  5. if/kimi-k2-thinking       (gratuit illimité)
-
-Résultat : 5 niveaux de fallback = zéro temps d'arrêt
-```
-
-### Cas 4 : « Je veux l'IA GRATUITE dans OpenClaw »
-
-**Problème :** Besoin d'assistant IA dans les apps de messagerie, entièrement gratuit
-
-```
-Combo : "openclaw-free"
-  1. if/glm-4.7                (illimité gratuit)
-  2. if/minimax-m2.1           (illimité gratuit)
-  3. if/kimi-k2-thinking       (illimité gratuit)
-
-Coût mensuel : 0 $
-Accès via : WhatsApp, Telegram, Slack, Discord, iMessage, Signal...
-```
-
----
-
-## 💡 Fonctionnalités principales
-
-### 🧠 Routage & Intelligence
-
-| Fonctionnalité                        | Ce qu'elle fait                                                                 |
-| ------------------------------------- | ------------------------------------------------------------------------------- |
-| 🎯 **Fallback intelligent 4 niveaux** | Auto-routage : Abonnement → Clé API → Économique → Gratuit                      |
-| 📊 **Suivi des quotas en temps réel** | Comptage de tokens en direct + compte à rebours de réinitialisation             |
-| 🔄 **Traduction de format**           | OpenAI ↔ Claude ↔ Gemini ↔ Cursor ↔ Kiro transparent                            |
-| 👥 **Support multi-comptes**          | Plusieurs comptes par fournisseur avec sélection intelligente                   |
-| 🔄 **Renouvellement auto des tokens** | Les tokens OAuth se renouvellent automatiquement avec retry                     |
-| 🎨 **Combos personnalisés**           | 6 stratégies : fill-first, round-robin, p2c, random, least-used, cost-optimized |
-| 🧩 **Modèles personnalisés**          | Ajoutez n'importe quel ID de modèle à n'importe quel fournisseur                |
-| 🌐 **Routeur wildcard**               | Routez les patterns `provider/*` vers n'importe quel fournisseur dynamiquement  |
-| 🧠 **Budget de raisonnement**         | Modes passthrough, auto, custom et adaptive pour les modèles de raisonnement    |
-| 💬 **Injection System Prompt**        | System prompt global appliqué à toutes les requêtes                             |
-| 📄 **API Responses**                  | Support complet de l'API Responses d'OpenAI (`/v1/responses`) pour Codex        |
-
-### 🎵 APIs multi-modales
-
-| Fonctionnalité             | Ce qu'elle fait                                         |
-| -------------------------- | ------------------------------------------------------- |
-| 🖼️ **Génération d'images** | `/v1/images/generations` — 4 fournisseurs, 9+ modèles   |
-| 📐 **Embeddings**          | `/v1/embeddings` — 6 fournisseurs, 9+ modèles           |
-| 🎤 **Transcription audio** | `/v1/audio/transcriptions` — compatible Whisper         |
-| 🔊 **Texte vers parole**   | `/v1/audio/speech` — synthèse audio multi-fournisseur   |
-| 🛡️ **Modérations**         | `/v1/moderations` — vérifications de sécurité           |
-| 🔀 **Reranking**           | `/v1/rerank` — reclassement de pertinence des documents |
-
-### 🛡️ Résilience & Sécurité
-
-| Fonctionnalité                  | Ce qu'elle fait                                                      |
-| ------------------------------- | -------------------------------------------------------------------- |
-| 🔌 **Circuit Breaker**          | Ouverture/fermeture auto par fournisseur avec seuils configurables   |
-| 🛡️ **Anti-Thundering Herd**     | Mutex + sémaphore de rate-limit pour les fournisseurs avec clé API   |
-| 🧠 **Cache sémantique**         | Cache à deux niveaux (signature + sémantique) réduit coût et latence |
-| ⚡ **Idempotence des requêtes** | Fenêtre de dédup 5s pour les requêtes dupliquées                     |
-| 🔒 **Spoofing TLS Fingerprint** | Contournement de détection de bot via wreq-js                        |
-| 🌐 **Filtrage IP**              | Allowlist/blocklist pour le contrôle d'accès API                     |
-| 📊 **Rate limits éditables**    | RPM configurable, intervalle minimum, concurrence max                |
-
-### 📊 Observabilité & Analytique
-
-| Fonctionnalité                    | Ce qu'elle fait                                                           |
-| --------------------------------- | ------------------------------------------------------------------------- |
-| 📝 **Logs de requêtes**           | Mode debug avec logs complets requête/réponse                             |
-| 💾 **Logs SQLite**                | Logs proxy persistants survivant aux redémarrages                         |
-| 📊 **Tableau de bord analytique** | Recharts : cartes de stats, graphique d'utilisation, tableau fournisseurs |
-| 📈 **Suivi de progression**       | Événements SSE de progression opt-in pour le streaming                    |
-| 🧪 **Évaluations LLM**            | Tests avec golden set et 4 stratégies de correspondance                   |
-| 🔍 **Télémétrie des requêtes**    | Agrégation de latence p50/p95/p99 + traçage X-Request-Id                  |
-| 📋 **Logs + Quotas**              | Pages dédiées pour navigation des logs et suivi des quotas                |
-| 🏥 **Tableau de bord santé**      | Uptime, états circuit breaker, lockouts, stats cache                      |
-| 💰 **Suivi des coûts**            | Gestion de budget + configuration des prix par modèle                     |
-
-### ☁️ Déploiement & Synchronisation
-
-| Fonctionnalité                    | Ce qu'elle fait                                                                 |
-| --------------------------------- | ------------------------------------------------------------------------------- |
-| 💾 **Cloud Sync**                 | Synchroniser les paramètres entre appareils via Cloudflare Workers              |
-| 🌐 **Déployer partout**           | Localhost, VPS, Docker, Cloudflare Workers                                      |
-| 🔑 **Gestion des clés API**       | Générer, faire tourner et limiter les clés API par fournisseur                  |
-| 🧙 **Assistant de configuration** | Setup guidé en 4 étapes pour les nouveaux utilisateurs                          |
-| 🔧 **Tableau de bord CLI Tools**  | Configuration en un clic pour Claude, Codex, Cline, OpenClaw, Kilo, Antigravity |
-| 🔄 **Sauvegardes DB**             | Sauvegarde et restauration automatiques de tous les paramètres                  |
-
-<details>
-<summary><b>📖 Détails des fonctionnalités</b></summary>
-
-### 🎯 Fallback intelligent 4 niveaux
-
-Créez des combos avec fallback automatique :
-
-```
-Combo : "my-coding-stack"
-  1. cc/claude-opus-4-6        (votre abonnement)
-  2. nvidia/llama-3.3-70b      (API NVIDIA gratuite)
-  3. glm/glm-4.7               (backup économique, $0.6/1M)
-  4. if/kimi-k2-thinking       (fallback gratuit)
-
-→ Bascule automatiquement lorsque le quota est épuisé ou en cas d'erreurs
-```
-
-### 📊 Suivi des quotas en temps réel
-
-- Consommation de tokens par fournisseur
-- Compte à rebours de réinitialisation (5 heures, quotidien, hebdomadaire)
-- Estimation des coûts pour les niveaux payants
-- Rapports de dépenses mensuels
-
-### 🔄 Traduction de format
-
-Traduction transparente entre les formats :
-
-- **OpenAI** ↔ **Claude** ↔ **Gemini** ↔ **OpenAI Responses**
-- Votre CLI envoie le format OpenAI → OmniRoute traduit → Le fournisseur reçoit le format natif
-- Fonctionne avec tout outil supportant les endpoints OpenAI personnalisés
-
-### 👥 Support multi-comptes
-
-- Ajouter plusieurs comptes par fournisseur
-- Round-robin automatique ou routage par priorité
-- Basculement vers le compte suivant lorsqu'un quota est atteint
-
-### 🔄 Renouvellement automatique des tokens
-
-- Les tokens OAuth se renouvellent automatiquement avant expiration
-- Pas de réauthentification manuelle nécessaire
-- Expérience transparente sur tous les fournisseurs
-
-### 🎨 Combos personnalisés
-
-- Créer des combinaisons de modèles illimitées
-- 6 stratégies : fill-first, round-robin, power-of-two-choices, random, least-used, cost-optimized
-- Partager les combos entre appareils avec Cloud Sync
-
-### 🏥 Tableau de bord santé
-
-- Statut du système (uptime, version, utilisation mémoire)
-- États des circuit breakers par fournisseur (Closed/Open/Half-Open)
-- Statut des rate limits et lockouts actifs
-- Statistiques du cache de signatures
-- Télémétrie de latence (p50/p95/p99) + cache de prompt
-- Réinitialisation de la santé en un clic
-
-### 🔧 Playground du traducteur
-
-- Déboguer, tester et visualiser les traductions de format d'API
-- Envoyer des requêtes et voir comment OmniRoute traduit entre les formats des fournisseurs
-- Inestimable pour résoudre les problèmes d'intégration
-
-### 💾 Cloud Sync
-
-- Synchroniser fournisseurs, combos et paramètres entre appareils
-- Synchronisation en arrière-plan automatique
-- Stockage chiffré sécurisé
-
-</details>
-
----
-
-## 📖 Guide de configuration
-
-<details>
-<summary><b>💳 Fournisseurs par abonnement</b></summary>
-
-### Claude Code (Pro/Max)
-
-```bash
-Tableau de bord → Fournisseurs → Connecter Claude Code
-→ Connexion OAuth → Renouvellement auto des tokens
-→ Suivi de quota 5h + hebdomadaire
-
-Modèles :
-  cc/claude-opus-4-6
-  cc/claude-sonnet-4-5-20250929
-  cc/claude-haiku-4-5-20251001
-```
-
-**Conseil Pro :** Utilisez Opus pour les tâches complexes, Sonnet pour la vitesse. OmniRoute suit les quotas par modèle !
-
-### OpenAI Codex (Plus/Pro)
-
-```bash
-Tableau de bord → Fournisseurs → Connecter Codex
-→ Connexion OAuth (port 1455)
-→ Reset 5h + hebdomadaire
-
-Modèles :
-  cx/gpt-5.2-codex
-  cx/gpt-5.1-codex-max
-```
-
-### Gemini CLI (GRATUIT 180K/mois !)
-
-```bash
-Tableau de bord → Fournisseurs → Connecter Gemini CLI
-→ Google OAuth
-→ 180K completions/mois + 1K/jour
-
-Modèles :
-  gc/gemini-3-flash-preview
-  gc/gemini-2.5-pro
-```
-
-**Meilleure valeur :** Niveau gratuit énorme ! Utilisez avant les niveaux payants.
-
-### GitHub Copilot
-
-```bash
-Tableau de bord → Fournisseurs → Connecter GitHub
-→ OAuth via GitHub
-→ Reset mensuel (1er du mois)
-
-Modèles :
-  gh/gpt-5
-  gh/claude-4.5-sonnet
-  gh/gemini-3-pro
-```
-
-</details>
-
-<details>
-<summary><b>🔑 Fournisseurs par clé API</b></summary>
-
-### NVIDIA NIM (GRATUIT 1000 crédits !)
-
-1. S'inscrire : [build.nvidia.com](https://build.nvidia.com)
-2. Obtenir une clé API gratuite (1000 crédits d'inférence inclus)
-3. Tableau de bord → Ajouter fournisseur → NVIDIA NIM :
-   - API Key : `nvapi-your-key`
-
-**Modèles :** `nvidia/llama-3.3-70b-instruct`, `nvidia/mistral-7b-instruct` et 50+ autres
-
-**Conseil Pro :** API compatible OpenAI — fonctionne parfaitement avec la traduction de format d'OmniRoute !
-
-### DeepSeek
-
-1. S'inscrire : [platform.deepseek.com](https://platform.deepseek.com)
-2. Obtenir une clé API
-3. Tableau de bord → Ajouter fournisseur → DeepSeek
-
-**Modèles :** `deepseek/deepseek-chat`, `deepseek/deepseek-coder`
-
-### Groq (Niveau gratuit disponible !)
-
-1. S'inscrire : [console.groq.com](https://console.groq.com)
-2. Obtenir une clé API (niveau gratuit inclus)
-3. Tableau de bord → Ajouter fournisseur → Groq
-
-**Modèles :** `groq/llama-3.3-70b`, `groq/mixtral-8x7b`
-
-**Conseil Pro :** Inférence ultra-rapide — idéal pour le codage en temps réel !
-
-### OpenRouter (100+ modèles)
-
-1. S'inscrire : [openrouter.ai](https://openrouter.ai)
-2. Obtenir une clé API
-3. Tableau de bord → Ajouter fournisseur → OpenRouter
-
-**Modèles :** Accès à 100+ modèles de tous les grands fournisseurs via une seule clé API.
-
-</details>
-
-<details>
-<summary><b>💰 Fournisseurs économiques (Backup)</b></summary>
-
-### GLM-4.7 (Reset quotidien, $0.6/1M)
-
-1. S'inscrire : [Zhipu AI](https://open.bigmodel.cn/)
-2. Obtenir une clé API du Coding Plan
-3. Tableau de bord → Ajouter clé API :
-   - Fournisseur : `glm`
-   - API Key : `your-key`
-
-**Utilisez :** `glm/glm-4.7`
-
-**Conseil Pro :** Le Coding Plan offre 3× le quota à 1/7 du coût ! Reset quotidien à 10h.
-
-### MiniMax M2.1 (Reset 5h, $0.20/1M)
-
-1. S'inscrire : [MiniMax](https://www.minimax.io/)
-2. Obtenir une clé API
-3. Tableau de bord → Ajouter clé API
-
-**Utilisez :** `minimax/MiniMax-M2.1`
-
-**Conseil Pro :** L'option la moins chère pour le contexte long (1M tokens) !
-
-### Kimi K2 (9 $/mois fixe)
-
-1. S'abonner : [Moonshot AI](https://platform.moonshot.ai/)
-2. Obtenir une clé API
-3. Tableau de bord → Ajouter clé API
-
-**Utilisez :** `kimi/kimi-latest`
-
-**Conseil Pro :** 9 $/mois fixe pour 10M tokens = 0,90 $/1M de coût effectif !
-
-</details>
-
-<details>
-<summary><b>🆓 Fournisseurs GRATUITS (Backup d'urgence)</b></summary>
-
-### iFlow (8 modèles GRATUITS)
-
-```bash
-Tableau de bord → Connecter iFlow
-→ Connexion OAuth iFlow
-→ Utilisation illimitée
-
-Modèles :
-  if/kimi-k2-thinking
-  if/qwen3-coder-plus
-  if/glm-4.7
-  if/minimax-m2
-  if/deepseek-r1
-```
-
-### Qwen (3 modèles GRATUITS)
-
-```bash
-Tableau de bord → Connecter Qwen
-→ Autorisation par code d'appareil
-→ Utilisation illimitée
-
-Modèles :
-  qw/qwen3-coder-plus
-  qw/qwen3-coder-flash
-```
-
-### Kiro (Claude GRATUIT)
-
-```bash
-Tableau de bord → Connecter Kiro
-→ AWS Builder ID ou Google/GitHub
-→ Utilisation illimitée
-
-Modèles :
-  kr/claude-sonnet-4.5
-  kr/claude-haiku-4.5
-```
-
-</details>
-
-<details>
-<summary><b>🎨 Créer des combos</b></summary>
-
-### Exemple 1 : Maximiser l'abonnement → Backup économique
-
-```
-Tableau de bord → Combos → Créer nouveau
-
-Nom : premium-coding
-Modèles :
-  1. cc/claude-opus-4-6 (Abonnement principal)
-  2. glm/glm-4.7 (Backup économique, $0.6/1M)
-  3. minimax/MiniMax-M2.1 (Fallback le moins cher, $0.20/1M)
-
-Utilisez en CLI : premium-coding
-```
-
-### Exemple 2 : Gratuit uniquement (Zéro coût)
-
-```
-Nom : free-combo
-Modèles :
-  1. gc/gemini-3-flash-preview (180K gratuits/mois)
-  2. if/kimi-k2-thinking (illimité)
-  3. qw/qwen3-coder-plus (illimité)
-
-Coût : 0 $ pour toujours !
-```
-
-</details>
-
-<details>
-<summary><b>🔧 Intégration CLI</b></summary>
-
-### Cursor IDE
-
-```
-Paramètres → Modèles → Avancé :
-  OpenAI API Base URL : http://localhost:20128/v1
-  OpenAI API Key : [du tableau de bord OmniRoute]
-  Model : cc/claude-opus-4-6
-```
-
-### Claude Code
-
-Utilisez la page **CLI Tools** dans le tableau de bord pour la configuration en un clic, ou modifiez `~/.claude/settings.json` manuellement.
-
-### Codex CLI
-
-```bash
-export OPENAI_BASE_URL="http://localhost:20128"
-export OPENAI_API_KEY="your-omniroute-api-key"
-
-codex "your prompt"
-```
-
-### OpenClaw
-
-**Option 1 — Tableau de bord (recommandé) :**
-
-```
-Tableau de bord → CLI Tools → OpenClaw → Sélectionner modèle → Appliquer
-```
-
-**Option 2 — Manuel :** Modifier `~/.openclaw/openclaw.json` :
-
-```json
-{
-  "models": {
-    "providers": {
-      "omniroute": {
-        "baseUrl": "http://127.0.0.1:20128/v1",
-        "apiKey": "sk_omniroute",
-        "api": "openai-completions"
-      }
-    }
-  }
-}
-```
-
-> **Note :** OpenClaw fonctionne uniquement avec OmniRoute local. Utilisez `127.0.0.1` au lieu de `localhost` pour éviter les problèmes de résolution IPv6.
-
-### Cline / Continue / RooCode
-
-```
-Paramètres → Configuration API :
-  Fournisseur : OpenAI Compatible
-  Base URL : http://localhost:20128/v1
-  API Key : [du tableau de bord OmniRoute]
-  Model : if/kimi-k2-thinking
-```
-
-</details>
-
----
-
-## 📊 Modèles disponibles
-
-<details>
-<summary><b>Voir tous les modèles disponibles</b></summary>
-
-**Claude Code (`cc/`)** - Pro/Max :
-
-- `cc/claude-opus-4-6`
-- `cc/claude-sonnet-4-5-20250929`
-- `cc/claude-haiku-4-5-20251001`
-
-**Codex (`cx/`)** - Plus/Pro :
-
-- `cx/gpt-5.2-codex`
-- `cx/gpt-5.1-codex-max`
-
-**Gemini CLI (`gc/`)** - GRATUIT :
-
-- `gc/gemini-3-flash-preview`
-- `gc/gemini-2.5-pro`
-
-**GitHub Copilot (`gh/`)** :
-
-- `gh/gpt-5`
-- `gh/claude-4.5-sonnet`
-
-**NVIDIA NIM (`nvidia/`)** - Crédits GRATUITS :
-
-- `nvidia/llama-3.3-70b-instruct`
-- `nvidia/mistral-7b-instruct`
-- 50+ modèles sur [build.nvidia.com](https://build.nvidia.com)
-
-**GLM (`glm/`)** - $0.6/1M :
-
-- `glm/glm-4.7`
-
-**MiniMax (`minimax/`)** - $0.2/1M :
-
-- `minimax/MiniMax-M2.1`
-
-**iFlow (`if/`)** - GRATUIT :
-
-- `if/kimi-k2-thinking`
-- `if/qwen3-coder-plus`
-- `if/deepseek-r1`
-- `if/glm-4.7`
-- `if/minimax-m2`
-
-**Qwen (`qw/`)** - GRATUIT :
-
-- `qw/qwen3-coder-plus`
-- `qw/qwen3-coder-flash`
-
-**Kiro (`kr/`)** - GRATUIT :
-
-- `kr/claude-sonnet-4.5`
-- `kr/claude-haiku-4.5`
-
-**OpenRouter (`or/`)** - 100+ modèles :
-
-- `or/anthropic/claude-4-sonnet`
-- `or/google/gemini-2.5-pro`
-- Tout modèle de [openrouter.ai/models](https://openrouter.ai/models)
-
-</details>
-
----
-
-## 🧪 Évaluations (Evals)
-
-OmniRoute inclut un framework d'évaluation intégré pour tester la qualité des réponses LLM contre un golden set. Accès via **Analytics → Evals** dans le tableau de bord.
-
-### Golden Set intégré
-
-Le « OmniRoute Golden Set » préchargé contient 10 cas de test :
-
-- Salutations, mathématiques, géographie, génération de code
-- Conformité format JSON, traduction, markdown
-- Rejet de sécurité (contenu nocif), comptage, logique booléenne
-
-### Stratégies d'évaluation
-
-| Stratégie  | Description                                                    | Exemple                          |
-| ---------- | -------------------------------------------------------------- | -------------------------------- |
-| `exact`    | La sortie doit correspondre exactement                         | `"4"`                            |
-| `contains` | La sortie doit contenir la sous-chaîne (insensible à la casse) | `"Paris"`                        |
-| `regex`    | La sortie doit correspondre au motif regex                     | `"1.*2.*3"`                      |
-| `custom`   | Fonction JS personnalisée retourne true/false                  | `(output) => output.length > 10` |
-
----
-
-## 🐛 Dépannage
-
-<details>
-<summary><b>Cliquez pour développer le guide de dépannage</b></summary>
-
-**« Language model did not provide messages »**
-
-- Quota du fournisseur épuisé → Vérifiez le suivi de quota dans le tableau de bord
-- Solution : Utilisez un combo avec fallback ou passez à un niveau moins cher
-
-**Rate limiting**
-
-- Quota d'abonnement épuisé → Fallback vers GLM/MiniMax
-- Ajoutez un combo : `cc/claude-opus-4-6 → glm/glm-4.7 → if/kimi-k2-thinking`
-
-**Token OAuth expiré**
-
-- Renouvelé automatiquement par OmniRoute
-- Si le problème persiste : Tableau de bord → Fournisseur → Reconnecter
-
-**Coûts élevés**
-
-- Vérifiez les statistiques d'utilisation dans Tableau de bord → Coûts
-- Changez le modèle principal pour GLM/MiniMax
-- Utilisez le niveau gratuit (Gemini CLI, iFlow) pour les tâches non critiques
-
-**Le tableau de bord s'ouvre sur le mauvais port**
-
-- Définissez `PORT=20128` et `NEXT_PUBLIC_BASE_URL=http://localhost:20128`
-
-**Erreurs de cloud sync**
-
-- Vérifiez que `BASE_URL` pointe vers votre instance en cours d'exécution
-- Vérifiez que `CLOUD_URL` pointe vers le point de terminaison cloud attendu
-- Gardez les valeurs `NEXT_PUBLIC_*` alignées avec les valeurs du serveur
-
-**Le premier login ne fonctionne pas**
-
-- Vérifiez `INITIAL_PASSWORD` dans `.env`
-- Si non défini, le mot de passe par défaut est `123456`
-
-**Pas de logs de requêtes**
-
-- Définissez `ENABLE_REQUEST_LOGS=true` dans `.env`
-
-**Le test de connexion affiche « Invalid » pour les fournisseurs compatibles OpenAI**
-
-- Beaucoup de fournisseurs n'exposent pas le point de terminaison `/models`
-- OmniRoute v1.0.6+ inclut une validation de secours via chat completions
-- Assurez-vous que l'URL de base inclut le suffixe `/v1`
-
-</details>
-
----
-
-## 🛠️ Stack technologique
-
-- **Runtime** : Node.js 20+
-- **Langage** : TypeScript 5.9 — **100% TypeScript** dans `src/` et `open-sse/` (v1.0.6)
-- **Framework** : Next.js 16 + React 19 + Tailwind CSS 4
-- **Base de données** : LowDB (JSON) + SQLite (état du domaine + logs proxy)
-- **Streaming** : Server-Sent Events (SSE)
-- **Auth** : OAuth 2.0 (PKCE) + JWT + API Keys
-- **Tests** : Node.js test runner (368+ tests unitaires)
-- **CI/CD** : GitHub Actions (publication automatique npm + Docker Hub lors du release)
-- **Site web** : [omniroute.online](https://omniroute.online)
-- **Package** : [npmjs.com/package/omniroute](https://www.npmjs.com/package/omniroute)
-- **Docker** : [hub.docker.com/r/diegosouzapw/omniroute](https://hub.docker.com/r/diegosouzapw/omniroute)
-- **Résilience** : Circuit breaker, backoff exponentiel, anti-thundering herd, spoofing TLS
-
----
-
-## 📖 Documentation
-
-| Document                                   | Description                                         |
-| ------------------------------------------ | --------------------------------------------------- |
-| [Guide utilisateur](docs/USER_GUIDE.md)    | Fournisseurs, combos, intégration CLI, déploiement  |
-| [Référence API](docs/API_REFERENCE.md)     | Tous les endpoints avec exemples                    |
-| [Dépannage](docs/TROUBLESHOOTING.md)       | Problèmes courants et solutions                     |
-| [Architecture](docs/ARCHITECTURE.md)       | Architecture système et détails internes            |
-| [Contribuer](CONTRIBUTING.md)              | Configuration de développement et directives        |
-| [Spécification OpenAPI](docs/openapi.yaml) | Spécification OpenAPI 3.0                           |
-| [Politique de sécurité](SECURITY.md)       | Signalement de vulnérabilités et pratiques sécurité |
-
----
-
-## 📧 Support
-
-> 💬 **Rejoignez notre communauté !** [Groupe WhatsApp](https://chat.whatsapp.com/JI7cDQ1GyaiDHhVBpLxf8b?mode=gi_t) — Obtenez de l'aide, partagez des astuces et restez informé.
-
-- **Site web** : [omniroute.online](https://omniroute.online)
-- **GitHub** : [github.com/diegosouzapw/OmniRoute](https://github.com/diegosouzapw/OmniRoute)
-- **Issues** : [github.com/diegosouzapw/OmniRoute/issues](https://github.com/diegosouzapw/OmniRoute/issues)
-- **WhatsApp** : [Groupe communautaire](https://chat.whatsapp.com/JI7cDQ1GyaiDHhVBpLxf8b?mode=gi_t)
-- **Projet original** : [9router par decolua](https://github.com/decolua/9router)
-
----
-
-## 👥 Contributeurs
-
-[![Contributors](https://contrib.rocks/image?repo=diegosouzapw/OmniRoute&max=100&columns=20&anon=1)](https://github.com/diegosouzapw/OmniRoute/graphs/contributors)
-
-### Comment contribuer
-
-1. Forkez le dépôt
-2. Créez votre branche de fonctionnalité (`git checkout -b feature/amazing-feature`)
-3. Committez vos changements (`git commit -m 'Add amazing feature'`)
-4. Poussez vers la branche (`git push origin feature/amazing-feature`)
-5. Ouvrez une Pull Request
-
-Consultez [CONTRIBUTING.md](CONTRIBUTING.md) pour les directives détaillées.
-
-### Publier une nouvelle version
-
-```bash
-# Créer un release — la publication npm est automatique
-gh release create v1.0.6 --title "v1.0.6" --generate-notes
-```
-
----
-
-## 📊 Historique des Stars
-
-<a href="https://star-history.com/#diegosouzapw/OmniRoute&Date">
- <picture>
-   <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/svg?repos=diegosouzapw/OmniRoute&type=Date&theme=dark" />
-   <source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/svg?repos=diegosouzapw/OmniRoute&type=Date" />
-   <img alt="Star History Chart" src="https://api.star-history.com/svg?repos=diegosouzapw/OmniRoute&type=Date" />
- </picture>
-</a>
-
----
-
-## 🙏 Remerciements
-
-Remerciements spéciaux à **[9router](https://github.com/decolua/9router)** par **[decolua](https://github.com/decolua)** — le projet original qui a inspiré ce fork. OmniRoute construit sur cette base incroyable avec des fonctionnalités supplémentaires, des APIs multi-modales et une réécriture complète en TypeScript.
-
-Remerciements spéciaux à **[CLIProxyAPI](https://github.com/router-for-me/CLIProxyAPI)** — l'implémentation originale en Go qui a inspiré ce portage en JavaScript.
-
----
-
-## 📄 Licence
-
-Licence MIT — voir [LICENSE](LICENSE) pour les détails.
-
----
-
-<div align="center">
-  <sub>Fait avec ❤️ pour les développeurs qui codent 24/7</sub>
-  <br/>
-  <sub><a href="https://omniroute.online">omniroute.online</a></sub>
-</div>