fix(docker): copy open-sse workspace manifest before npm ci (v3.8.39 image build) (#5223)

* chore(release): open v3.8.39 development cycle

* docs(changelog): backfill 5 v3.8.38 bullets merged after release finalize

These PRs squash-merged into release/v3.8.38 between the CHANGELOG finalize
(ff57be32f) and the merge-to-main (ae6e2342d), so they shipped in the v3.8.38
tag but had no bullet:

- feat(compression): Ionizer engine (lossy JSON-array sampling + CCR) (#5148)
- fix(sse): preserve non-stream reasoning fields (#5155, @rdself)
- fix(i18n): add missing English UI labels (#5153, @rdself)
- test(combo): gated live smoke (#5151) + release-expectations refresh (#5150, @KooshaPari)

(#5129 exact-host Anthropic baseUrl is already covered by the #5130 bullet — same CodeQL #674.)
Synced 41 i18n CHANGELOG mirrors.

* feat(compression): TOON best-of-N candidate encoder + encoder A/B table (#5163)

Integrated into release/v3.8.39. TOON best-of-N candidate encoder (GCF default, fail-open). 17/17 unit tests pass on merge result; CI reds were base-stale + Quality Ratchet DRIFT.

* fix(zenmux): normalize vendor-prefixed GLM system roles (#5158)

Integrated into release/v3.8.39. ZenMux vendor-prefixed GLM system-role normalization; 12/12 role-normalizer tests pass on merge result. CI reds base-stale.

* [codex] fix xAI OAuth test and reasoning effort (#5157)

Integrated into release/v3.8.39. xAI reasoning-effort normalization (max/xhigh→high) + OAuth test config; 46/46 xai-translator tests pass on merge result. CI reds base-stale.

* docs(i18n): add Traditional Chinese (zh-TW) README and update zh-CN to latest (#5162)

Integrated into release/v3.8.39. Traditional Chinese (zh-TW) README + zh-CN refresh; docs-only.

* test(security): guard PII redaction stays opt-in (default off) + Hard Rule #20 (#5159)

Integrated into release/v3.8.39. PII opt-in regression guard + Hard Rule #20; rebased to strip base-drift (+81/-1). 5/5 guard tests pass; flip-proof verified.

* test(combo): deterministic context-relay universal-handoff coverage (closes phase-2 TODO) (#5168)

Integrated into release/v3.8.39. Deterministic context-relay universal-handoff coverage (3 tests); 3/3 pass on merge result.

* docs(i18n): full sync zh-TW and zh-CN README with canonical English v3.8.39 (#5171)

Integrated into release/v3.8.39. Full zh-TW docs tree + zh-CN sync with canonical English v3.8.39; docs-only.

* fix(serve): honour HOSTNAME from .env instead of hardcoding 0.0.0.0 (#5134) (#5170)

Integrated into release/v3.8.39. HOSTNAME env override in serve (#5134) + regression test (4/4, TDD flip-proof verified).

* fix(sse): resolve nameless deepseek-web tool blocks via parameter-schema match (#5154) (#5173)

Integrated into release/v3.8.39. Schema-based nameless deepseek-web tool-block resolution (#5154); 6/6 tests pass on merge result (incl. ambiguous/no-match negatives + named-tag no-regression).

* fix(sse): normalize array user content for Command Code to avoid upstream 400 (#5166) (#5174)

Integrated into release/v3.8.39. Normalize array user content for Command Code (#5166, user-array/400 symptom); 4/4 tests pass on merge result.

* fix(sse): defer </think> close so it never leaks before tool_calls (#5123) (#5175)

Integrated into release/v3.8.39. Defer </think> close so it never leaks before tool_calls (#5123); 4/4 tests pass (incl. #4633 no-regression). CHANGELOG synced to keep all 3 v3.8.39 fixes.

* fix(dashboard): use amber for home update-step warning icon (#5176)

Integrated into release/v3.8.39. Amber for home update-step warning icon; 1/1 UI test.

* fix(api): LAN/Tailscale dashboard — host-aware CSP + GET-exempt version route + combo field errors (#5083) (#5177)

Integrated into release/v3.8.39. Host-aware CSP (ReDoS/injection-safe host validation) + GET-exempt /api/system/version (POST/spawn stays LOCAL_ONLY, exact-match safe-methods-only) + COMBO_002 firstField. 44/44 tests + route-guard membership gate green. CHANGELOG synced to keep all 4 v3.8.39 fixes.

* fix(api): replace #5083 global middleware CSP with declarative ws: scheme (#5083)

Follow-up to PR #5177 (merged): that version implemented the LAN-CSP fix (Bug 1)
with a new global `src/middleware.ts` + `src/server/csp.ts`, which contradicts the
project's documented architecture — 'No global Next.js middleware — interception is
route-specific' (CLAUDE.md / AGENTS.md) — and was merged unverified (middleware vs
next.config header precedence was never confirmed in a real build).

This replaces that approach with the minimal, declarative equivalent:
  • next.config.mjs: connect-src now permits the bare `ws:` scheme (symmetric with the
    bare `wss:` already allowed) so the dashboard can reach its own Live WS server from
    a LAN/Tailscale host. No middleware.
  • Removes src/middleware.ts, src/server/csp.ts, and tests/unit/csp-host-aware.test.ts.
  • Adds tests/unit/csp-lan-ws-5083.test.ts (incl. a guard asserting src/middleware.ts
    does NOT exist, so the global-middleware approach cannot silently return).

Bugs 2 (GET-exempt /api/system/version) and 3 (COMBO_002 field surfacing) from #5177
are unaffected and remain in place.

Co-authored-by: KooshaPari <KooshaPari@users.noreply.github.com>

* test(combo): end-to-end quota-share DRR routing-decision coverage (matrix parity) (#5179)

Integrated into release/v3.8.39. Quota-share DRR routing-decision coverage (matrix parity); 2/2 pass on merge result.

* feat(agent-bridge): graceful cert-install fallback with manual guide for containers (#4546) (#5178)

Integrated into release/v3.8.39. Agent-bridge graceful cert-install fallback + manual guide (#4546); 6/6 tests pass on merge result.

* fix(antigravity): family-scoped quota lockout (gemini/claude buckets) (#5180)

Integrated into release/v3.8.39 — family-scoped antigravity quota lockout. Rebased from v3.8.37 + validated (vitest 5/5, typecheck clean, full combo-matrix green, model-lockout 99/0). Same-model cross-account retry (chat.ts) deferred pending live antigravity VPS validation.

* fix(cli): force NODE_ENV to match dev/start run mode in custom Next server (#5189)

Integrated into release/v3.8.39. Force NODE_ENV to match dev/start run mode in custom Next server; 2/2 source-scan+ordering tests pass on merge result.

* feat(compression): CCR ranged/grep/stats retrieval (ReDoS-safe, backward-compat) (#5187)

Integrated into release/v3.8.39. CCR ranged/grep/stats retrieval (safe-regex ReDoS guard + length/match caps); 17/17 tests pass on merge result.

* docs(combo): sync all combo/routing-strategy docs to current state + document test coverage (#5185)

Integrated into release/v3.8.39. Combo/routing-strategy docs sync; docs-only.

* fix(mcp): return 404 (not 400) for unknown Streamable HTTP session id (#5169) (#5191)

* fix(api): respect blocked Auto (Zero-Config) provider in /v1/models catalog (#5192) (#5194)

* test(combo): deterministic context-relay codex quota-handoff coverage (closes last gap) (#5195)

* test(ci): wire antigravity-quota-family under test:vitest (fix test-discovery orphan) (#5196)

* fix(oauth): antigravity login no longer hangs — fire-and-forget onboarding + bounded post-exchange (#5193)

Antigravity OAuth hang fix (no-PKCE/no-openid + bounded post-exchange + exchange-500 fix). Includes #5200 (Koosha) revert + owner rebaseline to keep documented comments. Integrated into release/v3.8.39.

* feat(oauth): remote Antigravity login via local helper + paste-credentials (#5203)

Remote Antigravity login: local helper (omniroute login antigravity) + paste-credentials. Integrated into release/v3.8.39.

* fix(translator): accept Claude Messages shape in non-stream malformed-200 guard (#5156)

Integrated into release/v3.8.39

* fix(cli): default dev bundler to Turbopack (16.2.x panic no longer reproduces) (#5206)

Integrated into release/v3.8.39

* fix(cli): auto-calibrate server V8 heap from physical RAM (#5172) (#5213)

The server was spawned with a fixed --max-old-space-size=512 (omniroute serve)
or no heap flag at all (Electron), so RAM-rich boxes still OOM-crashed under
load (Ineffective mark-compacts near heap limit ~500MB) with many providers/
accounts and large model catalogs. New calibrateHeapFallbackMb(os.totalmem())
defaults the heap to ~35% of RAM clamped [512,4096], wired into serve.mjs and
electron/main.js. Explicit OMNIROUTE_MEMORY_MB still wins (#2939 unchanged).

Also addresses #5160 (same OOM root); #5152 (docker) benefits via the same knob.

Closes #5172

* fix(proxy): coalesce fast-fail health probes (#5208)

Integrated into release/v3.8.39

* fix(proxy): close dispatchers when clearing cache (#5202)

Integrated into release/v3.8.39

* fix(cli): raise dev server Node heap limit to 8GB to prevent OOM (#5198)

Integrated into release/v3.8.39

* fix(auth): allow synthetic no-auth fallback for mimocode (#5205)

Integrated into release/v3.8.39

* fix(oauth): preserve Antigravity refresh_token on empty/omitted upstream response (#3850) (#5214)

Google's OAuth refresh tokens are non-rotating: the refresh response usually
omits refresh_token and occasionally returns it as an empty string. The
Antigravity executor used `typeof tokens.refresh_token === "string" ? ... `
which accepts "" (typeof "" === "string") and overwrote the stored token with
empty, nulling it on first refresh. Now treats non-string OR empty as absent and
preserves credentials.refreshToken, matching refreshGoogleToken semantics.

Closes #3850

* fix(responses): normalize non-array input (#5204)

Integrated into release/v3.8.39

* fix(stream): normalize safety finish reasons via shared helper (#5197)

Integrated into release/v3.8.39

* fix(request-logger): never render negative '(-100%)' compression badge (#5201)

Integrated into release/v3.8.39

* fix(combo): reject empty responses api output (#5207)

Integrated into release/v3.8.39 — combo failover now rejects empty Responses API output (validateQuality). Baseline rebaseline dropped (main-measured drift; maintainer rebaselines at release).

* fix(pwa): prefer cached navigation before offline page (#5209)

Integrated into release/v3.8.39 — PWA service worker prefers cached navigation before offline page (#5165).

* chore(release): v3.8.39 — 2026-06-28

* chore(release): rebaseline openapi+i18n coverage ratchet drift for v3.8.39

---------

Co-authored-by: Arthur Bodera <abodera@gmail.com>
Co-authored-by: Nguyen Minh <lop123thcs@gmail.com>
Co-authored-by: lunkerchen <labanchen@gmail.com>
Co-authored-by: Ankit <177378174+anki1kr@users.noreply.github.com>
Co-authored-by: KooshaPari <KooshaPari@users.noreply.github.com>
Co-authored-by: Ardem2025 <ardemb22@gmail.com>
Co-authored-by: backryun <bakryun0718@proton.me>
Co-authored-by: Anton <39598727+NomenAK@users.noreply.github.com>
Co-authored-by: KooshaPari <42529354+KooshaPari@users.noreply.github.com>
Co-authored-by: Wilson <pedbookmed@gmail.com>
Co-authored-by: Randi <55005611+rdself@users.noreply.github.com>

.agents/workflows/capture-release-evidences.md

@@ -1,49 +0,0 @@
----
-description: Automatically run the browser_subagent to visually validate all new UI features from the current release and capture evidence WebP recordings of the changes.
----
-
-# Capture Release Evidences Workflow
-
-Use this workflow to automatically drive the `browser_subagent` to explore the newly deployed or locally running application and record evidence of the UI changes introduced in the latest release.
-
-## Prerequisites
-
-- OmniRoute must be actively running and accessible (e.g. locally at `http://localhost:20128` or on the Local VPS at `http://192.168.0.15:20128`).
-- The user must provide the target URL to be tested, or default to `http://192.168.0.15:20128`.
-
-## Workflow Steps
-
-### 1. Identify Target Features
-
-Review the `CHANGELOG.md` for the latest version to map out the new UI elements. For example:
-
-- **CLI Tools Settings**
-- **New Provider/Model Listings (e.g., Gemini 3.1, Qoder PAT)**
-- **New Feature Modals**
-
-### 2. Run the Browser Subagent
-
-For each identified feature, invoke the `browser_subagent` using the `default_api:browser_subagent` tool.
-**Important Task Guidelines for the Subagent:**
-
-- `TaskName`: Give it a clear name like "Validate CLIProxyAPI Tool Tab".
-- `TaskSummary`: "Navigate to the CLI Tools tab and verify the new Integration settings."
-- `Task`: Provide unambiguous instructions for the subagent, such as: "Navigate to http://192.168.0.15:20128/dashboard. Click on the 'Settings' or 'CLI Tools' nav link. Scroll down to find the CLIProxyAPI integration card. Hover over it to trigger UI state. Verify the components render correctly and exit."
-- `RecordingName`: Ensure it describes the feature (e.g. `v3_4_5_cli_proxy_api`). This is required and strictly automatically saved as a WebP artifacts video by the system.
-
-_(Note: The `browser_subagent` automatically creates a WebP recording named by the `RecordingName` parameter. No additional tools for screenshots are needed.)_
-
-### 3. Generate Report Artifact
-
-After the `browser_subagent` finishes its sessions, generate a final Markdown artifact (using `write_to_file` and `IsArtifact=true`) to present the recordings inline to the user using the `![caption](/absolute/path/to/media.webp)` syntax.
-
-### Example Invocation
-
-\```json
-{
-"TaskName": "Validating Qoder PAT Configuration UI",
-"TaskSummary": "Validates the Qoder provider configuration modal",
-"Task": "Go to http://192.168.0.15:20128/dashboard. Click on the 'Providers' tab. Find 'Qoder' in the list. Click 'Add Token' or 'Configure'. Type 'test_token' and submit. Return when done.",
-"RecordingName": "qoder_pat_ui_validation"
-}
-\```

.agents/workflows/deploy-vps-akamai.md

@@ -1,39 +0,0 @@
----
-description: Deploy the latest OmniRoute code to the Akamai VPS (69.164.221.35)
----
-
-# Deploy to Akamai VPS Workflow
-
-Deploy OmniRoute to the Akamai VPS using `npm pack + scp` + PM2.
-
-**Akamai VPS:** `69.164.221.35`
-**Process manager:** PM2 (`omniroute`)
-**Port:** `20128`
-
-## Steps
-
-### 1. Build + pack locally
-
-// turbo
-
-```bash
-cd /home/diegosouzapw/dev/proxys/9router && rm -f omniroute-*.tgz && rm -rf .next/cache app/.next/cache && npm run build:cli && rm -rf app/logs app/coverage app/.git app/.app-build-backup* && npm pack --ignore-scripts
-```
-
-### 2. Copy to Akamai VPS and install
-
-// turbo-all
-
-```bash
-scp omniroute-*.tgz root@69.164.221.35:/tmp/
-```
-
-```bash
-ssh root@69.164.221.35 "npm install -g /tmp/omniroute-*.tgz --ignore-scripts && cd /usr/lib/node_modules/omniroute/app && npm rebuild better-sqlite3 && pm2 delete omniroute 2>/dev/null; pm2 start /root/.omniroute/ecosystem.config.cjs --update-env && pm2 save && echo '✅ Akamai done'"
-```
-
-### 3. Verify the deployment
-
-```bash
-curl -s -o /dev/null -w 'AKAMAI HTTP %{http_code}\n' http://69.164.221.35:20128/
-```

.agents/workflows/deploy-vps-both.md

@@ -1,49 +0,0 @@
----
-description: Deploy the latest OmniRoute code to BOTH the Akamai VPS and the Local VPS
----
-
-# Deploy to VPS (Both) Workflow
-
-Deploy OmniRoute to the production VPSs using `npm pack + scp` + PM2.
-
-**Akamai VPS:** `69.164.221.35`
-**Local VPS:** `192.168.0.15`
-**Process manager:** PM2 (`omniroute`)
-**Port:** `20128`
-**PM2 entry:** `/usr/lib/node_modules/omniroute/app/server.js`
-
-> [!IMPORTANT]
-> The npm registry rejects packages > 100MB, so deployment uses **npm pack + scp**.
-
-## Steps
-
-### 1. Build + pack locally
-
-// turbo
-
-```bash
-cd /home/diegosouzapw/dev/proxys/9router && rm -f omniroute-*.tgz && rm -rf .next/cache app/.next/cache && npm run build:cli && rm -rf app/logs app/coverage app/.git app/.app-build-backup* && npm pack --ignore-scripts
-```
-
-### 2. Copy to both VPS and install
-
-// turbo-all
-
-```bash
-scp omniroute-*.tgz root@69.164.221.35:/tmp/ && scp omniroute-*.tgz root@192.168.0.15:/tmp/
-```
-
-```bash
-ssh root@69.164.221.35 "npm install -g /tmp/omniroute-*.tgz --ignore-scripts && cd /usr/lib/node_modules/omniroute/app && npm rebuild better-sqlite3 && pm2 delete omniroute 2>/dev/null; pm2 start /root/.omniroute/ecosystem.config.cjs --update-env && pm2 save && echo '✅ Akamai done'"
-```
-
-```bash
-ssh root@192.168.0.15 "npm install -g /tmp/omniroute-*.tgz --ignore-scripts && cd /usr/lib/node_modules/omniroute/app && npm rebuild better-sqlite3 && pm2 delete omniroute 2>/dev/null; pm2 start /root/.omniroute/ecosystem.config.cjs --update-env && pm2 save && echo '✅ Local done'"
-```
-
-### 3. Verify the deployment
-
-```bash
-curl -s -o /dev/null -w 'AKAMAI HTTP %{http_code}\n' http://69.164.221.35:20128/
-curl -s -o /dev/null -w 'LOCAL HTTP %{http_code}\n' http://192.168.0.15:20128/
-```

.agents/workflows/deploy-vps-local.md

@@ -1,39 +0,0 @@
----
-description: Deploy the latest OmniRoute code to the Local VPS (192.168.0.15)
----
-
-# Deploy to Local VPS Workflow
-
-Deploy OmniRoute to the Local VPS using `npm pack + scp` + PM2.
-
-**Local VPS:** `192.168.0.15`
-**Process manager:** PM2 (`omniroute`)
-**Port:** `20128`
-
-## Steps
-
-### 1. Build + pack locally
-
-// turbo
-
-```bash
-cd /home/diegosouzapw/dev/proxys/9router && rm -f omniroute-*.tgz && rm -rf .next/cache app/.next/cache && npm run build:cli && rm -rf app/logs app/coverage app/.git app/.app-build-backup* && npm pack --ignore-scripts
-```
-
-### 2. Copy to Local VPS and install
-
-// turbo-all
-
-```bash
-scp omniroute-*.tgz root@192.168.0.15:/tmp/
-```
-
-```bash
-ssh root@192.168.0.15 "npm install -g /tmp/omniroute-*.tgz --ignore-scripts && cd /usr/lib/node_modules/omniroute/app && npm rebuild better-sqlite3 && pm2 delete omniroute 2>/dev/null; pm2 start /root/.omniroute/ecosystem.config.cjs --update-env && pm2 save && echo '✅ Local done'"
-```
-
-### 3. Verify the deployment
-
-```bash
-curl -s -o /dev/null -w 'LOCAL HTTP %{http_code}\n' http://192.168.0.15:20128/
-```

.agents/workflows/generate-release.md

@@ -1,306 +0,0 @@
----
-description: Create a new release, bump version up to 1.x.10 threshold, update changelog, and manage Pull Requests
----
-
-# Generate Release Workflow
-
-Bump version, finalize CHANGELOG, commit, open a **PR to main** and wait for user confirmation before tagging, publishing, and deploying.
-
-> **VERSION RULE: Always use PATCH bumps (2.x.y → 2.x.y+1)**
-> NEVER use `npm version minor` or `npm version major`.
-> Always use: `npm version patch --no-git-tag-version`
-> The threshold rule: when `y` reaches 10, bump to `2.(x+1).0` — e.g. `2.1.10` → `2.2.0`.
-
-> **🔴 SINGLE BRANCH RULE**: The `release/vX.Y.Z` branch is the **ONLY** development branch for the entire release cycle. ALL work — bug fixes, feature implementations, PR integrations, issue resolutions — MUST be committed directly on this branch. Never create separate `fix/`, `feat/`, or topic branches. When running `/resolve-issues`, `/implement-features`, or `/review-prs`, always work on the current release branch.
-
----
-
-## ⚠️ Two-Phase Flow
-
-```
-Phase 1 (automated): bump → docs → i18n → commit → push → open PR
-  ↕  🛑 STOP: Notify user, wait for PR confirmation
-Phase 2 (post-merge): tag → publish → GitHub release → Docker → deploy
-```
-
-**NEVER push directly to main or create tags before the user confirms the PR.**
-
----
-
-## Phase 0: Security Verification (MANDATORY)
-
-Before creating the release, you must ensure the codebase and supply chain are secure and free of known vulnerabilities.
-
-1. **Run Local Dependencies Audit:**
-
-   ```bash
-   npm audit
-   ```
-
-   _Fix any `high` or `critical` vulnerabilities identified._
-
-2. **Check GitHub CodeQL & Dependabot Alerts:**
-   Navigate to the repository's **Security** tab on GitHub, or use the project's `vulnerability-scanner` skill to analyze active alerts. Ensure all static analysis findings (e.g., prototype pollution, insecure randomness, ReDoS, shell injections) are addressed and logically committed on a target branch.
-
----
-
-## Phase 1: Pre-Merge
-
-### 1. Create release branch
-
-```bash
-git checkout -b release/v2.x.y
-```
-
-### 2. Determine and sync version
-
-Check current version in `package.json`:
-
-```bash
-grep '"version"' package.json
-```
-
-> **🔴 BRANCH-VERSION PARITY RULE**: The logical version in `package.json` MUST exactly match the release branch name. For example, if you are on `release/v3.7.0`, the version in `package.json` MUST be `3.7.0`.
->
-> - If this is the FIRST time generating a release for a new minor/major branch (e.g., bumping from 3.6.9 to 3.7.0), you MUST ensure the version is bumped to match the new branch logic.
-> - If you are just bumping a patch on the current branch (e.g., 3.6.9 to 3.6.10), use:
->   `npm version patch --no-git-tag-version`
-
-> **⚠️ ATOMIC COMMIT RULE — Version bump MUST happen before committing feature files.**
->
-> **CORRECT order:**
->
-> 1. `npm version patch --no-git-tag-version` ← bump first
-> 2. implement features / fix bugs
-> 3. `git add -A && git commit -m "chore(release): v2.x.y — all changes in ONE commit"`
->
-> **OR if features are already staged:**
->
-> 1. implement features (do NOT commit yet)
-> 2. `npm version patch --no-git-tag-version` ← bump before committing
-> 3. `git add -A && git commit -m "chore(release): v2.x.y — all changes in ONE commit"`
->
-> **NEVER do this (creates version mismatch in git history):**
->
-> - ~~commit features → then bump version → commit package.json separately~~
->
-> This ensures that `git show v2.x.y` always contains both code changes and the version bump together.
-> The GitHub release tag will point to a commit that includes ALL changes for that version.
-
-### 3. Regenerate lock file (REQUIRED after version bump)
-
-**Mandatory** — skipping causes `@swc/helpers` lock mismatch and CI failures:
-
-```bash
-npm install
-```
-
-### 4. Finalize CHANGELOG.md
-
-> **🔴 NO MIXUPS RULE**: Ensure you do NOT mix the backlog of the previous version with the new one. The new version section must ONLY contain the features and fixes for the current release.
-
-Replace the `[Unreleased]` header with the new version and date.
-Keep an empty `## [Unreleased]` section above it, separated by a horizontal rule (`---`).
-
-```markdown
-## [Unreleased]
-
----
-
-## [3.7.0] — 2026-04-19
-
-### ✨ New Features
-
-- ...
-
-### 🐛 Bug Fixes
-
-- ...
-
----
-
-## [3.6.9] — 2026-04-19
-```
-
-### 5. Update openapi.yaml version ⚠️ MANDATORY
-
-> **CI will fail** if `docs/openapi.yaml` version ≠ `package.json` version (`check:docs-sync` enforces this).
-
-// turbo
-
-```bash
-VERSION=$(node -p "require('./package.json').version")
-sed -i "s/  version: .*/  version: $VERSION/" docs/openapi.yaml
-echo "✓ openapi.yaml → $VERSION"
-
-for dir in electron open-sse; do
-  if [ -d "$dir" ] && [ -f "$dir/package.json" ]; then
-    (cd "$dir" && npm version "$VERSION" --no-git-tag-version --allow-same-version > /dev/null)
-    echo "✓ $dir/package.json → $VERSION"
-  fi
-done
-# Re-run install to assert the workspace lockfile is updated
-npm install
-```
-
-### 6. Update README.md and i18n docs
-
-Run `/update-docs` workflow steps to:
-
-- Update feature table rows in `README.md`
-- Sync changes to all 29 language `docs/i18n/*/README.md` files
-- Update `docs/FEATURES.md` if Settings section changed
-
-### 7. Run tests
-
-// turbo
-
-```bash
-npm test
-```
-
-All tests must pass before creating the PR.
-
-### 8. Stage, commit, and push
-
-// turbo-all
-
-```bash
-git add -A
-git commit -m "chore(release): v2.x.y — summary of changes"
-git push origin release/v2.x.y
-```
-
-### 9. Open PR to main
-
-### 9. Open PR to main
-
-// turbo
-
-```bash
-VERSION=$(node -p "require('./package.json').version")
-
-# Extract the exact changelog entry for this version from the root CHANGELOG.md
-awk "/^## \\[$VERSION\\]/{flag=1; print; next} /^---/{if(flag) {flag=0; exit}} flag" CHANGELOG.md > /tmp/changelog_body.txt
-
-# Append test status and next steps
-echo "" >> /tmp/changelog_body.txt
-echo "### Tests" >> /tmp/changelog_body.txt
-echo "- All tests pass" >> /tmp/changelog_body.txt
-echo "" >> /tmp/changelog_body.txt
-echo "### ⚠️ After merging: run Phase 2 steps to tag, publish, and deploy." >> /tmp/changelog_body.txt
-
-gh pr create \
-  --repo diegosouzapw/OmniRoute \
-  --base main \
-  --head release/v$VERSION \
-  --title "Release v$VERSION" \
-  --body-file /tmp/changelog_body.txt
-```
-
-### 10. 🛑 STOP — Notify User & Await PR Confirmation
-
-**This is a mandatory stop point.** Use `notify_user` with `BlockedOnUser: true`:
-
-Inform the user:
-
-- PR URL
-- Summary of changes
-- Test results
-- List of files changed
-
-**DO NOT proceed to Phase 2 until the user confirms the PR looks good and merges it.**
-
----
-
-## Phase 2: Post-Merge (only after user confirms)
-
-> Run these steps only AFTER the user has merged the PR.
-
-### 11. Create Git Tag and GitHub Release (MANDATORY)
-
-// turbo
-
-```bash
-git checkout main
-git pull origin main
-VERSION=$(node -p "require('./package.json').version")
-
-# Extracts the changelog section for this version
-NOTES=$(awk '/^## \['"$VERSION"'\]/{flag=1; next} /^## \[[0-9]+/{if(flag) exit} flag' CHANGELOG.md | sed -e 's/^[[:space:]]*//' -e 's/[[:space:]]*$//')
-if [ -z "$NOTES" ]; then NOTES="OmniRoute v$VERSION Release"; fi
-
-git tag -a "v$VERSION" -m "Release v$VERSION"
-git push origin --tags
-gh release create "v$VERSION" --title "v$VERSION" --notes "$NOTES" --target main
-```
-
-### 14. 🐳 Trigger Docker Hub build (MANDATORY — keep npm and Docker in sync)
-
-> **CRITICAL**: Docker Hub and npm MUST always publish the same version.
-> The Docker image is built automatically via GitHub Actions when a new tag is pushed.
-> After pushing the tag in step 11-12, **verify the workflow runs**:
-
-```bash
-# Verify the Docker workflow triggered
-gh run list --repo diegosouzapw/OmniRoute --workflow docker-publish.yml --limit 3
-
-# Wait for the Docker build to complete (usually 5–10 min)
-gh run watch --repo diegosouzapw/OmniRoute
-
-# After completion, verify on Docker Hub:
-# https://hub.docker.com/r/diegosouzapw/omniroute/tags
-```
-
-If the Docker build was not triggered automatically, trigger it manually:
-
-```bash
-gh workflow run docker-publish.yml --repo diegosouzapw/OmniRoute --ref v2.x.y
-```
-
-### 15. Deploy to BOTH VPS environments (MANDATORY)
-
-> Always deploy to **both** environments after every release.
-> See `/deploy-vps` workflow for detailed steps.
-
-```bash
-# Build and pack locally
-cd /home/diegosouzapw/dev/proxys/9router && rm -f omniroute-*.tgz && rm -rf .next/cache app/.next/cache && npm run build:cli && rm -rf app/logs app/coverage app/.git app/.app-build-backup* && npm pack --ignore-scripts
-
-# Deploy to LOCAL VPS (192.168.0.15)
-scp omniroute-*.tgz root@192.168.0.15:/tmp/
-ssh root@192.168.0.15 "npm install -g /tmp/omniroute-*.tgz --ignore-scripts && cd /usr/lib/node_modules/omniroute/app && npm rebuild better-sqlite3 && pm2 delete omniroute 2>/dev/null; pm2 start /root/.omniroute/ecosystem.config.cjs --update-env && pm2 save && echo '✅ Local done'"
-
-# Deploy to AKAMAI VPS (69.164.221.35)
-scp omniroute-*.tgz root@69.164.221.35:/tmp/
-ssh root@69.164.221.35 "npm install -g /tmp/omniroute-*.tgz --ignore-scripts && cd /usr/lib/node_modules/omniroute/app && npm rebuild better-sqlite3 && pm2 delete omniroute 2>/dev/null; pm2 start /root/.omniroute/ecosystem.config.cjs --update-env && pm2 save && echo '✅ Akamai done'"
-
-# Verify both
-curl -s -o /dev/null -w "LOCAL:  HTTP %{http_code}\n" http://192.168.0.15:20128/
-curl -s -o /dev/null -w "AKAMAI: HTTP %{http_code}\n" http://69.164.221.35:20128/
-```
-
-### 16. Preserve release branch
-
-```bash
-# Branch is kept for historical purposes. Do not delete.
-```
-
----
-
-## Notes
-
-- Always run `/update-docs` BEFORE this workflow (ensures CHANGELOG and README are current)
-- The `prepublishOnly` script runs `npm run build:cli` automatically during `npm publish`
-- After npm publish, verify with `npm info omniroute version`
-- Lock file sync errors are caused by skipping `npm install` after version bump
-- Use `gh auth switch -u diegosouzapw` if git push fails with wrong account
-
-## Known CI Pitfalls
-
-| CI failure                                                                | Cause                                                    | Fix                                                                    |
-| ------------------------------------------------------------------------- | -------------------------------------------------------- | ---------------------------------------------------------------------- |
-| `[docs-sync] FAIL - OpenAPI version differs from package.json`            | Skipped step 5 — `docs/openapi.yaml` version not updated | Run step 5 (`sed -i ...`) and commit                                   |
-| `[docs-sync] FAIL - CHANGELOG.md first section must be "## [Unreleased]"` | `## [Unreleased]` missing or not at top of CHANGELOG     | Add `## [Unreleased]\n\n---\n` before the first versioned `## [x.y.z]` |
-| Electron Linux `.deb` build fails (`FpmTarget` error)                     | `fpm` Ruby gem not installed on `ubuntu-latest` runner   | Already fixed in `electron-release.yml` (`gem install fpm` step)       |
-| Docker Hub `502 error writing layer blob`                                 | Transient Docker Hub network error during ARM64 push     | Re-run the Docker publish workflow; no code change needed              |

.agents/workflows/implement-features.md

@@ -1,706 +0,0 @@
----
-description: Analyze open feature request issues, implement viable ones on dedicated branches, and respond to authors
----
-
-# /implement-features — Feature Request Harvest, Research & Implementation Workflow
-
-## Overview
-
-A **5-phase** workflow that systematically harvests feature requests from GitHub issues, creates structured idea files, researches solutions across the internet and Git repositories, presents a consolidated report for user approval, then generates detailed implementation plans and executes them.
-
-**Output directory structure:**
-
-```
-_ideia/
-├── viable/                  # Features approved for implementation
-│   ├── need_details/        # ❓ Good idea but waiting for author clarification (issues stay OPEN)
-│   │   └── 1015-warp-terminal-mitm.md
-│   ├── 1046-native-playground.md           # ✅ Ready — researched and planned
-│   └── 1046-native-playground.requirements.md
-├── defer/                   # ⏭️ Good ideas deferred for future cycles (issues CLOSED)
-│   └── 1041-smart-auto-combos.md
-└── notfit/                  # ❌ Out of scope / already exists (issues CLOSED)
-    └── 945-telegram-integration.md
-
-_tasks/features-vX.Y.Z/   # Implementation plans (per-release)
-└── 1046-native-playground.plan.md
-```
-
-> **LIFECYCLE RULE:** `viable/` files are **DELETED** once the feature is implemented — they are not moved. Only unimplemented features live in `viable/` (or `viable/need_details/`). Files in `defer/` and `notfit/` remain as permanent reference.
-
-> **BRANCH RULE**: All implementation work MUST happen on the current `release/vX.Y.Z` branch. Never create separate `feat/` branches. If no release branch exists yet, create one first using `/generate-release` Phase 1 steps 1–5.
-
----
-
-## Phase 1 — Harvest: Collect & Catalog Feature Ideas
-
-### 1.1 Identify the Repository
-
-// turbo
-
-- Run: `git -C <project_root> remote get-url origin` to extract owner/repo.
-
-### 1.2 Ensure Release Branch Exists
-
-// turbo
-
-Before doing any work, ensure you are on the current release branch:
-
-```bash
-# Check current branch
-git branch --show-current
-
-# If on main, determine next version and create the release branch
-VERSION=$(node -p "require('./package.json').version")
-NEXT=$(node -p "const [a,b,c]=('$VERSION').split('.').map(Number); c>=9?a+'.'+(b+1)+'.0':a+'.'+b+'.'+(c+1)")
-git checkout -b release/v$NEXT
-npm version patch --no-git-tag-version
-npm install
-```
-
-If already on a `release/vX.Y.Z` branch, continue working there.
-
-### 1.3 Fetch ALL Open Feature Requests
-
-// turbo-all
-
-**⚠️ CRITICAL**: The JSON output of `gh issue list` can be truncated by the tool, silently hiding issues. You MUST use the two-step approach below.
-
-**Step 1 — Get Issue numbers only** (small output, never truncated):
-
-```bash
-# Fetch issues with feature/enhancement labels
-gh issue list --repo <owner>/<repo> --state open -l "enhancement" --limit 500 --json number --jq '.[].number'
-
-# Also check for [Feature] in title (common pattern when no labels are set)
-gh issue list --repo <owner>/<repo> --state open --limit 500 --json number,title --jq '.[] | select(.title | test("\\[Feature\\]|\\[feature\\]|feature request"; "i")) | .number'
-```
-
-- Merge both lists, deduplicate. Count and confirm the total.
-
-**Step 2 — Fetch full metadata for each Issue** (one call per issue):
-
-```bash
-gh issue view <NUMBER> --repo <owner>/<repo> --json number,title,labels,body,comments,createdAt,author,assignees
-```
-
-- Read the **entire body** — including description, use cases, screenshots, mockups, and any embedded images.
-- Read **ALL comments** — community discussion, agreements, restrictions, owner responses, and linked PRs.
-- **Images**: If the body or comments contain image URLs (`![...](...)` or `https://...png/jpg/gif`), note them — they may contain UI mockups, wireframes, or architecture diagrams that are essential to understanding the request.
-- You may batch these into parallel calls (up to 4 at a time).
-- Sort by oldest first (FIFO).
-
-### 1.4 Create Idea Files (initially in `_ideia/` root)
-
-For each feature request, create a structured idea file in `<project_root>/_ideia/`:
-
-**Filename convention**: `<NUMBER>-<kebab-case-short-title>.md`
-Example: `1046-native-playground.md`, `1041-smart-auto-combos.md`
-
-#### 1.4a — If the idea file does NOT exist yet, create it:
-
-```markdown
-# Feature: <Title from Issue>
-
-> GitHub Issue: #<NUMBER> — opened by @<author> on <date>
-> Status: 📋 Cataloged | Priority: TBD
-
-## 📝 Original Request
-
-<Paste the FULL issue body here, preserving all formatting, images, and code blocks>
-
-## 💬 Community Discussion
-
-<Summarize ALL comments chronologically, noting who said what and any decisions or objections raised>
-
-### Participants
-
-- @<author> — Original requester
-- @<commenter1> — <brief role/opinion>
-- ...
-
-### Key Points
-
-- <bullet list of the most important discussion points>
-- <agreements reached>
-- <objections raised>
-
-## 🎯 Refined Feature Description
-
-<YOUR interpretation and enrichment of the feature request. Expand on what was asked, fill in logical gaps, provide concrete examples of how it would work. This section should be MORE detailed and clearer than the original request.>
-
-### What it solves
-
-- <problem 1>
-- <problem 2>
-
-### How it should work (high level)
-
-1. <step 1>
-2. <step 2>
-3. ...
-
-### Affected areas
-
-- <list of codebase areas, modules, files likely affected>
-
-## 📎 Attachments & References
-
-- <any image URLs, mockup links, or external references from the issue>
-
-## 🔗 Related Ideas
-
-- <links to related \_ideia/ files if any overlap found>
-```
-
-#### 1.4b — If the idea file ALREADY exists, update it:
-
-- Append new comments from the issue to the **Community Discussion** section.
-- Update the **Refined Feature Description** if new information changes the understanding.
-- Add any new **Related Ideas** cross-references found.
-- **Do NOT overwrite** existing content — append and enrich it.
-
-### 1.5 Cross-Reference & Deduplication
-
-After processing all issues:
-
-- Scan all `_ideia/*.md` files for overlapping features.
-- If two features are substantially the same, add `🔗 Related Ideas` cross-references to both.
-- If one is a strict subset of another, note it in the smaller file: `> ℹ️ This feature is a subset of #<OTHER_NUMBER>. Consider implementing together.`
-
----
-
-## Phase 2 — Research: Find Solutions & Build Requirements
-
-For each cataloged idea that is **viable** (aligns with the project's goals):
-
-### 2.1 Viability Pre-Check
-
-Before investing in research, quickly assess:
-
-- [ ] Does this feature align with the project's goals and architecture?
-- [ ] Is it technically feasible with the current codebase?
-- [ ] Does it duplicate existing functionality?
-- [ ] Would it introduce breaking changes or security risks?
-- [ ] Is there enough detail to understand what's needed?
-
-**Verdict options:**
-
-| Verdict               | When                                  | Action                      |
-| --------------------- | ------------------------------------- | --------------------------- |
-| ✅ **VIABLE**         | Good idea, enough context             | Proceed to Research         |
-| ❓ **NEEDS DETAIL**   | Good idea, insufficient spec          | Skip research, ask author   |
-| ⏭️ **DEFER**          | Good idea, too complex for this cycle | Catalog only, skip research |
-| ❌ **NOT FIT**        | Doesn't fit the project               | Explain why                 |
-| 🔁 **ALREADY EXISTS** | Feature already implemented           | Point to existing feature   |
-
-### 2.2 Internet Research (for VIABLE features)
-
-For each viable feature, perform systematic research:
-
-**Step 1 — Web search for similar implementations:**
-
-```
-search_web("how to implement <feature description> in <tech stack>")
-search_web("<feature keyword> implementation nextjs typescript 2025 2026")
-search_web("<feature keyword> open source library npm")
-```
-
-**Step 2 — Find reference Git repositories:**
-
-```
-search_web("site:github.com <feature keyword> <tech stack> stars:>100")
-search_web("github <feature keyword> implementation recently updated 2026")
-```
-
-- Find **up to 10 relevant repositories**, sorted by most recently updated.
-- For each repository:
-  - Note the repo URL, star count, last commit date
-  - Read its README and relevant source files via `read_url_content`
-  - Extract the architectural approach, patterns used, and key code snippets
-
-**Step 3 — Read API docs and standards:**
-
-If the feature involves an external API, protocol, or standard:
-
-- Find and read the official documentation
-- Note version requirements, authentication patterns, rate limits
-
-### 2.3 Create Requirements File
-
-For each researched feature, create a requirements file alongside its idea file:
-
-**Filename**: `<NUMBER>-<kebab-case-short-title>.requirements.md`
-
-```markdown
-# Requirements: <Feature Title>
-
-> Feature Idea: [#<NUMBER>](./<NUMBER>-<kebab-case-short-title>.md)
-> Research Date: <YYYY-MM-DD>
-> Verdict: ✅ VIABLE
-
-## 🔍 Research Summary
-
-<Brief summary of what was found during research>
-
-## 📚 Reference Implementations
-
-| #   | Repository       | Stars | Last Updated | Approach | Relevance    |
-| --- | ---------------- | ----- | ------------ | -------- | ------------ |
-| 1   | [repo/name](url) | ⭐ N  | YYYY-MM-DD   | <brief>  | High/Med/Low |
-| 2   | ...              |       |              |          |              |
-
-### Key Patterns Found
-
-- <pattern 1 with code snippet or link>
-- <pattern 2>
-
-## 📐 Proposed Solution Architecture
-
-### Approach
-
-<Describe the chosen approach based on research findings>
-
-### New Files
-
-| File                  | Purpose       |
-| --------------------- | ------------- |
-| `path/to/new/file.ts` | <description> |
-
-### Modified Files
-
-| File                       | Changes        |
-| -------------------------- | -------------- |
-| `path/to/existing/file.ts` | <what changes> |
-
-### Database Changes
-
-- <migrations needed, if any>
-
-### API Changes
-
-- <new/modified endpoints, if any>
-
-### UI Changes
-
-- <new/modified pages/components, if any>
-
-## ⚙️ Implementation Effort
-
-- **Estimated complexity**: Low / Medium / High / Very High
-- **Estimated files changed**: ~N
-- **Dependencies needed**: <new npm packages, if any>
-- **Breaking changes**: Yes/No — <details>
-- **i18n impact**: <number of new translation keys>
-- **Test coverage needed**: <brief description>
-
-## ⚠️ Open Questions
-
-- <question 1>
-- <question 2>
-
-## 🔗 External References
-
-- <documentation URLs>
-- <API references>
-```
-
----
-
-## Phase 2.5 — Organize & Respond: Sort Files and Post GitHub Comments
-
-### 2.5.1 Create Directory Structure
-
-// turbo
-
-```bash
-mkdir -p <project_root>/_ideia/viable
-mkdir -p <project_root>/_ideia/viable/need_details
-mkdir -p <project_root>/_ideia/defer
-mkdir -p <project_root>/_ideia/notfit
-```
-
-### 2.5.2 Move Idea Files to Category Subdirectories
-
-After classification, move EVERY idea file to its correct subdirectory:
-
-```bash
-# ✅ VIABLE — move idea + requirements files
-mv _ideia/<NUMBER>-*.md _ideia/viable/
-mv _ideia/<NUMBER>-*.requirements.md _ideia/viable/
-
-# ❓ NEEDS DETAIL — viable but waiting for author response
-mv _ideia/<NUMBER>-*.md _ideia/viable/need_details/
-
-# ⏭️ DEFER — move idea files only
-mv _ideia/<NUMBER>-*.md _ideia/defer/
-
-# ❌ NOT FIT & 🔁 ALREADY EXISTS — move idea files only
-mv _ideia/<NUMBER>-*.md _ideia/notfit/
-```
-
-No files should remain in `_ideia/` root after this step (except subdirectories).
-
-### 2.5.3 Post GitHub Comments by Category
-
-**Each category has a specific comment template and action:**
-
----
-
-#### For 🔁 ALREADY EXISTS — Comment + CLOSE issue
-
-// turbo
-
-The feature already exists in the system. Explain WHERE it is and HOW to use it.
-
-```markdown
-Hi @<author>! Thanks for the suggestion! 🙏
-
-Great news — this functionality **already exists** in OmniRoute:
-
-**📍 Where to find it:** <exact dashboard path or settings location>
-
-**🔧 How to use it:**
-
-1. <step 1>
-2. <step 2>
-3. <step 3>
-
-If you have any trouble finding or using it, feel free to ask in a Discussion. We're always happy to help!
-
-Closing this as the feature is already available. 🎉
-```
-
-```bash
-gh issue close <NUMBER> --repo <owner>/<repo> --comment "<comment above>"
-```
-
----
-
-#### For ⏭️ DEFER — Comment + CLOSE issue
-
-// turbo
-
-Thank the user, explain the idea was cataloged, and that we'll study it before implementing.
-
-```markdown
-Hi @<author>! Thanks for this thoughtful feature request! 🙏
-
-We really appreciate the detailed proposal. We've **cataloged your idea** and it's now part of our improvement backlog.
-
-Due to the **significant architectural impact** of this feature, we'll need to conduct thorough use-case studies and architectural analysis before we start development. This ensures we build it right and don't introduce regressions.
-
-**What happens next:**
-
-- Your idea is saved in our internal feature backlog
-- We'll conduct architecture studies when this area is prioritized
-- We'll notify you here when development begins
-
-Thank you for contributing to OmniRoute's roadmap! Your input helps shape the product. 🚀
-```
-
-```bash
-gh issue close <NUMBER> --repo <owner>/<repo> --comment "<comment above>"
-```
-
----
-
-#### For ❌ NOT FIT — Comment + CLOSE issue
-
-// turbo
-
-Politely explain why the feature doesn't fit the project scope.
-
-```markdown
-Hi @<author>! Thanks for the suggestion! 🙏
-
-After careful analysis, we've determined that this feature **falls outside OmniRoute's core scope** as a proxy/router.
-
-**Reason:** <explain why — e.g., "Telegram integration belongs in the application/orchestrator layer that consumes OmniRoute's API, not inside the router itself.">
-
-**Alternative:** <suggest an alternative approach if possible>
-
-We appreciate you thinking of ways to improve OmniRoute! If you'd like to discuss this further, feel free to open a Discussion. 🙏
-```
-
-```bash
-gh issue close <NUMBER> --repo <owner>/<repo> --comment "<comment above>"
-```
-
----
-
-#### For ❓ NEEDS DETAIL — Comment (keep OPEN)
-
-// turbo
-
-Ask for the specific missing details needed.
-
-```markdown
-Hi @<author>! Thanks for the feature request — it's an interesting idea and we'd love to explore it further. 🙏
-
-To move forward, we need a few more details:
-
-1. <specific question 1>
-2. <specific question 2>
-3. <specific question 3>
-
-If you know of any **open-source projects or repositories** that implement something similar, please share links — it would help us design the best solution.
-
-Looking forward to your response! 🚀
-```
-
----
-
-#### For ✅ VIABLE — Comment (keep OPEN)
-
-// turbo
-
-Thank the user, confirm we've cataloged their idea, and explain it may be implemented in future versions.
-
-```markdown
-Hi @<author>! Thanks for the great feature suggestion! 🙏
-
-We've analyzed your request and it aligns well with OmniRoute's roadmap. We've **cataloged this feature** and it's in our implementation backlog.
-
-**Status:** 📋 Cataloged for future implementation
-
-This feature may be included in upcoming releases. We'll **respond to this issue and tag you** as soon as implementation begins so you can test it.
-
-Thank you for helping improve OmniRoute! 🚀
-```
-
-**⚠️ Do NOT close viable issues — they remain OPEN for tracking.**
-
----
-
-## Phase 3 — Report: Present Findings to User
-
-### 3.1 🛑 MANDATORY STOP — Present Consolidated Report
-
-After completing Phase 1, Phase 2, and Phase 2.5, **STOP and present the following report** in the chat. Do NOT proceed to implementation.
-
-Present a structured report containing:
-
-#### 3.1a — Feature Summary Table
-
-| #   | Issue | Title | Verdict         | Location                      | Action                        |
-| --- | ----- | ----- | --------------- | ----------------------------- | ----------------------------- |
-| 1   | #N    | Title | ✅ VIABLE       | `_ideia/viable/`              | Issue OPEN, comment posted    |
-| 2   | #N    | Title | ⏭️ DEFER        | `_ideia/defer/`               | Issue CLOSED with explanation |
-| 3   | #N    | Title | ❌ NOT FIT      | `_ideia/notfit/`              | Issue CLOSED with explanation |
-| 4   | #N    | Title | 🔁 EXISTS       | `_ideia/notfit/`              | Issue CLOSED with guidance    |
-| 5   | #N    | Title | ❓ NEEDS DETAIL | `_ideia/viable/need_details/` | Issue OPEN, questions posted  |
-
-#### 3.1b — Viable Features Detail
-
-For each VIABLE feature, provide a brief paragraph:
-
-- What was found during research
-- The proposed approach
-- Key risks or unknowns
-- Which reference repositories were most useful
-
-#### 3.1c — Issues Requiring Author Feedback
-
-For features marked ❓ NEEDS DETAIL, list:
-
-- What specific information is missing
-- What examples or repository references would help
-
-#### 3.1d — Ask for User Confirmation
-
-End the report with:
-
-> **Ready to proceed with implementation?**
->
-> - Reply **"sim"** or **"yes"** to generate full implementation plans for all VIABLE features.
-> - Reply with specific issue numbers to select only certain features.
-> - Reply **"não"** or **"no"** to stop here.
-
----
-
-## Phase 4 — Plan: Generate Implementation Plans (after user says "yes")
-
-> **⚠️ Do NOT enter this phase without explicit user approval from Phase 3.**
-
-### 4.1 Create Task Directory
-
-```bash
-mkdir -p <project_root>/_tasks/features-vX.Y.Z/
-```
-
-### 4.2 Generate One Implementation Plan Per Feature
-
-For each VIABLE feature approved by the user, create:
-
-**Filename**: `_tasks/features-vX.Y.Z/<NUMBER>-<kebab-case-title>.plan.md`
-
-```markdown
-# Implementation Plan: <Feature Title>
-
-> Issue: #<NUMBER>
-> Idea: [\_ideia/viable/<NUMBER>-title.md](../../_ideia/viable/<NUMBER>-title.md)
-> Requirements: [\_ideia/viable/<NUMBER>-title.requirements.md](../../_ideia/viable/<NUMBER>-title.requirements.md)
-> Branch: `release/vX.Y.Z`
-
-## Overview
-
-<Brief description of what will be built>
-
-## Pre-Implementation Checklist
-
-- [ ] Read all related source files listed below
-- [ ] Confirm no conflicts with in-flight PRs
-- [ ] Verify database migration numbering
-
-## Implementation Steps
-
-### Step 1: <Title>
-
-**Files:**
-
-- `path/to/file.ts` — <what to change>
-
-**Details:**
-<Detailed description of the change, including code patterns to follow, function signatures, etc.>
-
-### Step 2: <Title>
-
-...
-
-### Step N: Tests
-
-**New test files:**
-
-- `tests/unit/<test-file>.test.mjs` — <what to test>
-
-**Test cases:**
-
-- [ ] <test case 1>
-- [ ] <test case 2>
-
-### Step N+1: i18n
-
-**Translation keys to add:**
-
-- `<namespace>.<key>` — "<English value>"
-
-### Step N+2: Documentation
-
-- [ ] Update CHANGELOG.md
-- [ ] Update relevant docs/ files
-
-## Verification Plan
-
-1. Run `npm run build` — must pass
-2. Run `npm test` — all tests must pass
-3. Run `npm run lint` — no new errors
-4. <Manual verification steps>
-
-## Commit Plan
-```
-
-feat: <description> (#<NUMBER>)
-
-```
-
-```
-
-### 4.3 Present Plans for Final Approval
-
-Present a summary of all generated plans:
-
-> **Implementation plans generated:**
->
-> | #   | Feature | Plan File                                | Steps   | Effort |
-> | --- | ------- | ---------------------------------------- | ------- | ------ |
-> | 1   | <title> | `_tasks/features-vX.Y.Z/N-title.plan.md` | N steps | Medium |
->
-> Reply **"sim"** or **"yes"** to begin implementation of all features.
-> Reply with specific issue numbers to implement only certain ones.
-
----
-
-## Phase 5 — Execute: Implement the Plans (after user says "yes")
-
-> **⚠️ Do NOT enter this phase without explicit user approval from Phase 4.**
-
-### 5.1 Implement Each Feature
-
-For each approved plan, execute it step by step:
-
-1. **Follow the plan** — implement exactly as specified in the `.plan.md` file
-2. **Build** — Run `npm run build` after each feature to verify compilation
-3. **Test** — Run `npm test` to ensure no regressions
-4. **Commit** — Commit with: `feat: <description> (#<NUMBER>)`
-5. **Update the plan** — Mark completed steps with `[x]` in the plan file
-6. **Continue** — Move to the next feature (do NOT switch branches)
-
-### 5.2 Respond to Authors (Update Viable Issues)
-
-For each implemented feature, **close the issue with a final comment**:
-
-````markdown
-✅ **Implemented in `release/vX.Y.Z`!**
-
-Hi @<author>! Great news — your feature request has been implemented! 🎉
-
-**What was done:**
-
-- <bullet list of what was built>
-
-**How to try it:**
-
-```bash
-git fetch origin && git checkout release/vX.Y.Z
-npm install && npm run dev
-```
-````
-
-This will be included in the upcoming **vX.Y.Z** release. Feel free to reopen if you spot any issues! 🚀
-
-````
-
-```bash
-gh issue close <NUMBER> --repo <owner>/<repo> --comment "<comment above>"
-````
-
-Then **DELETE the idea file** — it has served its purpose:
-
-```bash
-# ✅ Implemented files are DELETED (not moved)
-rm _ideia/viable/<NUMBER>-<title>.md
-rm _ideia/viable/<NUMBER>-<title>.requirements.md  # if exists
-```
-
-> **Why delete?** `viable/` only holds features that still NEED to be done. Once implemented, the commit history and CHANGELOG are the source of truth. Keeping the file would be confusing.
-
-### 5.3 Finalize & Push
-
-After implementing all approved features:
-
-1. **Update CHANGELOG.md** on the release branch with all new feature entries
-2. Push the release branch: `git push origin release/vX.Y.Z`
-3. Run `/generate-release` workflow Phase 1 steps 7–10 (tests → commit → push → open PR to main → wait for user)
-
-### 5.4 Final Summary Report
-
-Present a final summary report to the user:
-
-| Issue | Title | Verdict         | Action                                             | Commit    |
-| ----- | ----- | --------------- | -------------------------------------------------- | --------- |
-| #N    | Title | ✅ Implemented  | Issue closed, idea file deleted                    | `abc1234` |
-| #N    | Title | ⏭️ Deferred     | Issue closed + saved in `_ideia/defer/`            | —         |
-| #N    | Title | ❌ Not Fit      | Issue closed + saved in `_ideia/notfit/`           | —         |
-| #N    | Title | 🔁 Exists       | Issue closed + saved in `_ideia/notfit/`           | —         |
-| #N    | Title | ❓ Needs Detail | Issue OPEN, moved to `_ideia/viable/need_details/` | —         |
-
-Include:
-
-- Total features harvested
-- Total ideas cataloged (`viable/need_details/` + `defer/` + `notfit/`)
-- Total features implemented (idea files deleted, issues closed)
-- Total features deferred
-- Total issues closed
-- Total issues left open (needs detail only — viable are closed after implementation)
-- Test results (pass/fail count)

.agents/workflows/issue-triage.md

@@ -1,50 +0,0 @@
----
-description: How to respond to GitHub issues with insufficient information
----
-
-# Issue Triage Workflow
-
-Respond to GitHub issues that need more information before they can be investigated.
-
-## Steps
-
-### 1. Identify issues needing triage
-
-```bash
-gh issue list --state open --limit 20
-```
-
-### 2. Evaluate each issue
-
-Check if the issue has:
-
-- Clear reproduction steps
-- Environment details (OS, Node.js version, OmniRoute version)
-- Error logs/screenshots
-- Expected vs actual behavior
-
-### 3. Respond with triage template
-
-For issues missing information:
-
-```markdown
-Thank you for reporting this issue! To help us investigate, please provide:
-
-1. **OmniRoute version**: (`omniroute --version`)
-2. **Node.js version**: (`node --version`)
-3. **Operating system**: (e.g., Ubuntu 24.04, macOS 15, Windows 11)
-4. **Installation method**: (npm, Docker, source)
-5. **Steps to reproduce**: (exact commands/actions that trigger the issue)
-6. **Error logs**: (paste relevant logs from the console)
-7. **Expected behavior**: (what should happen)
-
-This will help us debug and resolve your issue faster. 🙏
-```
-
-### 4. Label the issue
-
-Add appropriate labels: `needs-info`, `bug`, `enhancement`, `question`, etc.
-
-```bash
-gh issue edit <NUMBER> --add-label "needs-info"
-```

.agents/workflows/resolve-issues.md

@@ -1,163 +0,0 @@
----
-description: Fetch all open GitHub issues, analyze bugs, resolve what's possible, triage the rest, wait for user validation, then commit and release
----
-
-# /resolve-issues — Automated Issue Resolution Workflow
-
-## Overview
-
-This workflow fetches all open issues from the project's GitHub repository, classifies them, analyzes bugs, proposes a resolution plan, waits for user validation, and ONLY THEN implements the fixes, commits, and closes the issues on the current release branch (`release/vX.Y.Z`). It does NOT merge or release automatically — the release branch is later merged via PR to main.
-
-> **BRANCH RULE**: All work MUST happen on the current `release/vX.Y.Z` branch. Never create separate `fix/` branches. If no release branch exists yet, create one first using `/generate-release` Phase 1 steps 1–5.
-
-## Steps
-
-### 1. Identify the GitHub Repository
-
-// turbo
-
-- Run: `git -C <project_root> remote get-url origin` to extract the owner/repo
-- Parse the owner and repo name from the URL
-
-### 2. Ensure Release Branch Exists
-
-// turbo
-
-Before doing any work, ensure you are on the current release branch:
-
-```bash
-# Check current branch
-git branch --show-current
-
-# If on main, determine next version and create the release branch
-VERSION=$(node -p "require('./package.json').version")
-NEXT=$(node -p "const [a,b,c]=('$VERSION').split('.').map(Number); c>=9?a+'.'+(b+1)+'.0':a+'.'+b+'.'+(c+1)")
-git checkout -b release/v$NEXT
-npm version patch --no-git-tag-version
-npm install
-```
-
-If already on a `release/vX.Y.Z` branch, continue working there.
-
-### 3. Fetch All Open Issues
-
-// turbo-all
-
-**⚠️ CRITICAL**: The JSON output of `gh issue list` can be truncated by the tool, silently hiding issues. You MUST use the two-step approach below to guarantee **all** issues are fetched.
-
-**Step 3a — Get Issue numbers only** (small output, never truncated):
-
-- Run: `gh issue list --repo <owner>/<repo> --state open --limit 500 --json number --jq '.[].number'`
-- This outputs one issue number per line. Count them and confirm total.
-
-**Step 3b — Fetch full metadata for each Issue** (one call per issue):
-
-- For each issue number from step 3a, run:
-  `gh issue view <NUMBER> --repo <owner>/<repo> --json number,title,labels,body,comments,createdAt,author`
-- You may batch these into parallel calls (up to 4 at a time).
-- Sort by oldest first (FIFO).
-
-### 4. Classify Each Issue
-
-For each issue, determine its type:
-
-- **Bug** — Has `bug` label, or body contains error messages, stack traces, "doesn't work", "broken", "crash", "error"
-- **Feature Request** — Has `enhancement`/`feature` label, or body describes new functionality
-- **Question** — Has `question` label, or is asking "how to" something
-- **Other** — Anything else
-
-Focus ONLY on **Bugs** for resolution. Feature requests and questions should be skipped with a note in the final report.
-
-### 5. Deep-Read Each Bug Issue (One-by-One Analysis)
-
-**IMPORTANT**: Read each bug issue thoroughly, one at a time, before moving to the next. This is NOT a batch process — each issue needs focused attention.
-
-#### 5a. Understand the Problem
-
-For each bug issue, perform the full analysis:
-
-1. **Read the entire body** — including Description, Steps to Reproduce, Expected/Actual Behavior, Error Logs, and Screenshots
-2. **Read ALL comments** — including bot triage comments (Kilo, etc.) and owner/community responses. Pay attention to:
-   - Whether someone already responded with a fix
-   - Whether a community member confirmed the issue is resolved
-   - Whether the issue was marked as duplicate by a bot
-3. **Identify the claimed error** — extract the exact error message, status code, and provider/model involved
-
-#### 5b. Check Information Sufficiency
-
-Verify the issue contains enough to act on:
-
-- [ ] Clear description of the problem
-- [ ] Steps to reproduce OR error logs
-- [ ] Provider/model/version information
-- [ ] Expected vs actual behavior
-
-#### 5c. Determine Issue Disposition
-
-For each bug, classify into one of 5 actions:
-
-| Disposition                  | When to Apply                                                                               | Action                                                  |
-| ---------------------------- | ------------------------------------------------------------------------------------------- | ------------------------------------------------------- |
-| **✅ CLOSE — Already Fixed** | Owner responded with fix + no user follow-up, OR community confirmed fix                    | Close with comment citing which version fixed it        |
-| **✅ CLOSE — Duplicate**     | Bot flagged >85% similarity + user provides no new info                                     | Close referencing the original issue                    |
-| **✅ CLOSE — Stale**         | We requested logs/info > 7 days ago with no reply                                           | Close thanking the user, invite to reopen if needed     |
-| **📝 RESPOND — Needs Info**  | Issue is real but missing critical reproduction details                                     | Comment asking for specifics per `/issue-triage`        |
-| **📝 RESPOND — User Config** | Error is caused by unsupported env (Node version, wrong model path, missing API enablement) | Comment explaining the user-side fix                    |
-| **🔧 FIX — Code Change**     | Root cause is confirmed in the codebase                                                     | Research, propose solution in report, wait for approval |
-
-#### 5d. For "FIX — Code Change" Issues
-
-Before coding, perform deep source analysis to formulate a plan:
-
-1. **Search the codebase** — `grep_search` for error strings, relevant function names, affected files
-2. **Search the web** — for upstream API changes, SDK updates, or breaking changes that explain the bug
-3. **Read the full source file** — don't rely on grep snippets; understand the surrounding logic
-4. **Verify the root cause** — confirm the bug is reproducible based on the code, not just a user misconfiguration
-5. **Formulate a proposed solution** — detail the exact files and lines you will change and how you will solve it.
-6. **DO NOT modify the codebase yet** — wait for user approval on your report first.
-
-#### 5e. For "RESPOND" Issues
-
-Post a substantive comment that:
-
-- Acknowledges the specific error they reported
-- Explains the likely root cause
-- Provides concrete steps to resolve (version upgrade, env var fix, model path correction)
-- Asks for follow-up info if needed
-
-**Do NOT post generic template responses.** Every comment should reference the user's specific error messages and environment.
-
-### 6. Generate Report & Wait for Validation
-
-Present a summary report to the user detailing your proposed actions. For any bugs that need fixing, explicitly explain your proposed solution (files to change and logic) and point out that it will be implemented on the release branch (`release/vX.Y.Z`) after approval.
-
-| Issue | Title | Status        | Proposed Action / Version                 |
-| ----- | ----- | ------------- | ----------------------------------------- |
-| #N    | Title | ✅ Close      | Already fixed / duplicate (explain why)   |
-| #N    | Title | 🔧 Propose    | Explanation of the code fix to be applied |
-| #N    | Title | 📝 Respond    | Guidance comment to be posted             |
-| #N    | Title | ❓ Needs Info | Triage comment to be posted               |
-| #N    | Title | ⏭️ Skip       | Feature request / not a bug               |
-
-> **⚠️ IMPORTANT**: Do NOT implement code changes, commit, push, or close issues at this step.
-> Wait for the user to review the proposed fixes and respond with **OK** before proceeding.
-
-- If the user says **OK** or approves → Proceed to step 7
-- If the user requests changes → Adjust the proposed solution and present the report again
-- If the user rejects → Revert any accidental changes and stop
-
-### 7. Implement Fixes, Run Tests & Commit (only after user approval)
-
-After the user validates and gives the OK:
-
-1. **Implement the fixes** — modify the codebase according to the approved plan.
-2. **Run tests** — `npm run test:all` (or the specific test file) to ensure 100% pass.
-3. **Update CHANGELOG.md** with all new bug fix entries.
-4. **Commit** each fix individually on the release branch with message format: `fix: <description> (#<issue_number>)`.
-5. **Push** the release branch: `git push origin release/vX.Y.Z`.
-6. **Close resolved issues immediately**. For each issue that was marked as Fixed, run:
-   `gh issue close <NUMBER> --repo <owner>/<repo> --comment "Thank you for reporting! This issue has been fixed and will be included in the next release (vX.Y.Z)."`
-7. Likewise, close `Duplicate` issues referencing the original, close `Needs Info` if stale, and post the required comments.
-8. If the project runs automatic releases or needs a PR, proceed to run `/generate-release` workflow Phase 1 steps 7–10 (tests → commit → push → open PR to main → wait for user).
-
-If NO fixes were committed, skip closing and source control steps and just conclude the workflow.

.agents/workflows/review-discussions.md

@@ -1,118 +0,0 @@
----
-description: Read all open GitHub Discussions, summarize them, respond to pending ones, and create issues from actionable feature requests
----
-
-# /review-discussions — GitHub Discussions Review & Response Workflow
-
-## Overview
-
-This workflow reads all open GitHub Discussions, generates a categorized summary, identifies which ones need a response, drafts and posts replies, and optionally creates issues from actionable feature requests. It follows the same flow used for Issues but adapted for the Discussions forum.
-
-// turbo-all
-
-## Steps
-
-### 1. Identify the GitHub Repository
-
-- Run: `git -C <project_root> remote get-url origin` to extract the owner/repo
-- Parse the owner and repo name from the URL
-
-### 2. Fetch All Open Discussions
-
-- Use `read_url_content` to fetch `https://github.com/<owner>/<repo>/discussions`
-- Parse the discussion list to get all discussion titles, IDs, authors, categories, and dates
-- For each discussion, fetch the individual page to read the full content and all comments/replies
-
-### 3. Summarize All Discussions
-
-For each discussion, extract:
-
-- **Title** and **#Number**
-- **Author** (GitHub username)
-- **Category** (Announcements, General, Ideas, Q&A, Show and tell)
-- **Date** created
-- **Summary** of the original post (1-2 sentences)
-- **Comments count** and key participants
-- **Your previous response** (if any)
-- **Pending action** — whether a response or follow-up is needed
-
-### 4. Present Summary Report to User
-
-Present the full summary to the user organized by category, using a table:
-
-| #   | Category | Title | Author | Date   | Status            |
-| --- | -------- | ----- | ------ | ------ | ----------------- |
-| #N  | Ideas    | Title | @user  | Mar 23 | ⚠️ Needs response |
-| #N  | Q&A      | Title | @user  | Mar 9  | ✅ Answered       |
-| #N  | General  | Title | @user  | Mar 19 | ⚠️ Needs response |
-
-Highlight:
-
-- **⚠️ Needs response** — No reply from maintainer, or a follow-up comment was left unanswered
-- **✅ Answered** — Maintainer already responded
-- **🐛 Bug reported** — A bug was mentioned that needs tracking
-- **💡 Actionable** — Contains a concrete feature request that could become an issue
-
-### 5. Draft & Post Responses
-
-For each discussion that needs a response, draft a reply following these guidelines:
-
-#### Response Style
-
-- **Friendly and professional** — Start with "Hey @username!"
-- **Acknowledge the contribution** — Thank the user for their input
-- **Be specific** — Reference existing features, settings, or dashboard pages if the feature already exists
-- **Provide workarounds** — If the request isn't implemented yet, suggest current alternatives
-- **Commit to action** — If the request is valid, state that you'll open an issue or add it to the roadmap
-- **Keep it concise** — 3-5 paragraphs max
-
-#### Posting via Browser
-
-- Use `browser_subagent` to navigate to each discussion and post the comment
-- **IMPORTANT**: When typing text in GitHub comment boxes via the browser, use only plain ASCII characters:
-  - Use regular hyphens `-` instead of em-dashes
-  - Use `->` instead of arrow symbols
-  - Do NOT use emoji Unicode characters (the browser keyboard may fail on them)
-  - Use `**bold**` and `\`code\`` markdown formatting
-- Click the green "Comment" button (or "Reply" for threaded replies) after typing
-- Verify the comment was posted by checking the page shows the new comment
-
-### 6. Create Issues from Actionable Feature Requests
-
-For discussions that contain concrete, actionable feature requests:
-
-1. Ask the user which ones should become issues
-2. For each approved request, create a GitHub issue via `browser_subagent`:
-   - Navigate to `https://github.com/<owner>/<repo>/issues/new`
-   - **Title**: `<Feature Name> - <Short description>`
-   - **Body** should include:
-     - `## Feature Request` header
-     - `**Source:** Discussion #N by @author`
-     - `## Problem` — What limitation the user hit
-     - `## Proposed Solution` — How it could work
-     - `### Implementation Ideas` — Technical approach
-     - `### Current Workarounds` — What users can do today
-     - `## Additional Context` — Links to related issues/discussions
-   - Add `enhancement` label
-   - Click "Submit new issue" / "Create"
-3. After creation, go back to the original discussion and post a comment linking to the new issue:
-   - "I've opened Issue #N to track this feature request. Follow along there for updates!"
-
-### 7. Final Report
-
-Present a final summary to the user:
-
-| Discussion | Action Taken                       |
-| ---------- | ---------------------------------- |
-| #N — Title | Responded with workarounds         |
-| #N — Title | Responded + created Issue #N       |
-| #N — Title | Already answered, no action needed |
-| #N — Title | Responded to follow-up comment     |
-
-## Notes
-
-- This workflow is **interactive** — always present the summary and wait for user approval before posting responses or creating issues
-- If the user says "pode responder" (or similar approval), proceed with posting all drafted responses
-- For discussions in non-English languages, respond in the same language as the original post
-- Always reference specific dashboard paths, config options, or code files when explaining existing features
-- When a discussion reveals a bug, note it separately from feature requests

.agents/workflows/review-prs.md

@@ -1,242 +0,0 @@
----
-description: Analyze open Pull Requests from the project's GitHub repository, generate a critical report, and optionally implement approved changes
----
-
-# /review-prs — PR Review & Analysis Workflow
-
-## Overview
-
-This workflow fetches all open PRs from the project's GitHub repository, performs a critical analysis of each one, generates a detailed report, and waits for user approval before proceeding with implementation. **All improvements are committed on the current release branch** (`release/vX.Y.Z`).
-
-> **BRANCH RULE**: PRs are ALWAYS merged into the current `release/vX.Y.Z` branch, NEVER directly into `main`. The release branch acts as a staging area — only after all PRs are integrated and tests pass does the release branch get merged into `main` via the `/generate-release` workflow.
-
-## Steps
-
-### 1. Identify the GitHub Repository
-
-- Read `package.json` to get the repository URL, or use the git remote origin URL
-  // turbo
-- Run: `git -C <project_root> remote get-url origin` to extract the owner/repo
-
-### 2. Ensure Release Branch Exists
-
-// turbo
-
-Before doing any work, ensure you are on the current release branch:
-
-```bash
-# Check current branch
-git branch --show-current
-
-# If on main, determine next version and create the release branch
-VERSION=$(node -p "require('./package.json').version")
-# Bump patch: e.g. 3.3.11 → 3.3.12
-NEXT=$(node -p "const [a,b,c]=('$VERSION').split('.').map(Number); c>=9?a+'.'+(b+1)+'.0':a+'.'+b+'.'+(c+1)")
-git checkout -b release/v$NEXT
-npm version patch --no-git-tag-version
-npm install
-```
-
-If already on a `release/vX.Y.Z` branch, continue working there.
-
-### 3. Fetch Open Pull Requests
-
-// turbo-all
-
-**⚠️ CRITICAL**: The JSON output of `gh pr list` can be truncated by the tool, silently hiding PRs. You MUST use the two-step approach below to guarantee **all** PRs are fetched.
-
-**Step 3a — Get PR numbers only** (small output, never truncated):
-
-- Run: `gh pr list --repo <owner>/<repo> --state open --limit 500 --json number --jq '.[].number'`
-- This outputs one PR number per line. Count them and confirm total.
-
-**Step 3b — Fetch full metadata for each PR** (one call per PR):
-
-- For each PR number from step 3a, run:
-  `gh pr view <NUMBER> --repo <owner>/<repo> --json number,title,author,headRefName,baseRefName,body,createdAt,additions,deletions,files`
-- You may batch these into parallel calls (up to 4 at a time).
-
-**Step 3c — Fetch diffs for each PR** (one call per PR, saved to /tmp):
-
-- For each PR number, run:
-  `gh pr diff <NUMBER> --repo <owner>/<repo> > /tmp/pr<NUMBER>.diff`
-- Then read each diff file with `view_file`.
-
-- For each open PR, collect:
-  - PR number, title, author, branch, number of commits, date
-  - PR description/body
-  - Files changed (diff)
-  - Existing review comments (from bots or humans)
-
-**Verification**: Confirm the count of PRs analyzed matches the count from step 3a before proceeding.
-
-### 3.5 Redirect PR Base Branches to Release Branch
-
-// turbo-all
-
-**⚠️ CRITICAL**: Contributors typically open PRs targeting `main`. Before analyzing or merging, redirect ALL open PRs to target the current release branch instead.
-
-```bash
-# Get the current release branch name
-RELEASE_BRANCH=$(git branch --show-current)  # e.g. release/v3.5.4
-
-# For each open PR that targets main, change its base to the release branch
-for PR_NUM in $(gh pr list --repo <owner>/<repo> --state open --json number,baseRefName --jq '.[] | select(.baseRefName == "main") | .number'); do
-  echo "Redirecting PR #$PR_NUM → $RELEASE_BRANCH"
-  gh pr edit "$PR_NUM" --repo <owner>/<repo> --base "$RELEASE_BRANCH"
-done
-```
-
-This ensures:
-
-1. PRs merge into the release branch, not directly into `main`
-2. Merge conflict detection is accurate against the release branch
-3. The release branch accumulates all changes before the final merge to `main`
-4. If the release branch doesn't exist on remote yet, push it first: `git push origin $RELEASE_BRANCH`
-
-### 4. Analyze Each PR — For each open PR, perform the following analysis:
-
-#### 4a. Feature Assessment
-
-- **Does it make sense?** Evaluate if the feature fills a real gap or solves a valid problem
-- **Alignment** — Check if it aligns with the project's architecture and roadmap
-- **Complexity** — Assess if the scope is reasonable or if it should be split
-
-#### 4b. Code Quality Review
-
-- Check for code duplication
-- Evaluate error handling patterns (consistent with existing codebase?)
-- Check naming conventions and code style
-- Verify TypeScript types (any `any` usage, missing types?)
-
-#### 4c. Security Review
-
-- Check for missing authentication/authorization on new endpoints
-- Check for injection vulnerabilities (URL params, SQL, XSS)
-- Verify input validation on all user-controlled data
-- Check for hardcoded secrets or credentials
-
-#### 4d. Architecture Review
-
-- Does the change follow existing patterns?
-- Are there any breaking changes to public APIs?
-- Is the database schema affected? Migration needed?
-- Impact on performance (N+1 queries, missing indexes?)
-
-#### 4e. Test Coverage
-
-- Does the PR include tests?
-- Are edge cases covered?
-- Would existing tests break?
-
-#### 4f. Cross-Layer (Global) Analysis
-
-Perform a **global impact assessment** to verify whether the PR changes are complete across all layers of the application:
-
-- **Backend → Frontend check**: If the PR adds or modifies backend-only resources (new endpoints, services, data models), evaluate whether corresponding frontend changes are missing:
-  - Does a new endpoint require a new screen/page in the dashboard?
-  - Should there be a new action button, menu item, or navigation link?
-  - Are there new data fields that should be displayed or editable in the UI?
-  - Does a new feature need a toggle, configuration panel, or status indicator?
-- **Frontend → Backend check**: If the PR adds frontend elements, verify the backend support exists:
-  - Are the required API endpoints implemented?
-  - Is the data model sufficient for the new UI components?
-- **Cross-cutting concerns**: Check shared layers (types, DTOs, validation schemas, routes, middleware) for completeness
-- **Document gaps** — If missing layers are detected, list them as **IMPORTANT** issues in the report with concrete suggestions for what should be added
-
-### 5. Generate Report — Create a markdown report for each PR including:
-
-- **PR Summary** — What it does, files affected, commit count
-- **Improvements/Benefits** — Numbered list with impact level (HIGH/MEDIUM/LOW)
-- **Risks & Issues** — Categorized as CRITICAL / IMPORTANT / MINOR
-- **Scoring Table** — Rate across: Feature Relevance, Code Quality, Security, Robustness, Tests
-- **Verdict** — Ready to merge? With mandatory vs optional fixes
-- **Next Steps** — What will happen if approved
-
-### 6. Present to User
-
-- Show the report via `notify_user` with `BlockedOnUser: true`
-- Wait for user decision:
-  - **Approved** → Proceed to step 7
-  - **Approved with changes** → Implement the fixes and corrections before merging
-  - **Rejected** → Close the PR or leave a review comment
-
-### 7. Pre-Merge Fixes & CI Green-Lighting (if approved)
-
-> **⚠️ Fixes and Conflict Resolutions MUST be pushed back to the PR branch before merging.** We want the PR itself to be green and fully valid before it integrates.
-
-- **Sync latest fixes & Resolve Conflicts:** Merge the current `release` branch into the PR branch. If there are merge conflicts, you MUST resolve them inside the author's PR branch. NEVER resolve conflicts by closing their PR and doing the work in a separate branch, as this steals credit from the original author.
-- **Implement improvements:** Apply the required fixes identified in the analysis directly on the PR branch (e.g., adding missing API routes, fixing SSRF, applying comments from other agents).
-- **Pushing changes to PR branches:**
-
-  ```bash
-  # Checkout the PR locally
-  gh pr checkout <NUMBER>
-
-  # Apply fixes, commit your changes
-  git commit -m "chore: apply review suggestions and missing layers"
-
-  # Attempt to push directly to the PR branch
-  git push
-  ```
-
-- **Fallback (For external forks without maintainer edit access):**
-  If `git push` fails because the PR comes from an external fork without write access, you MUST:
-  1. Create a new branch ending in `-fix` (e.g., `checkout -b fix-pr-<NUMBER>`).
-  2. Push your branch to the main repo (`git push origin fix-pr-<NUMBER>`).
-  3. Create a Pull Request targeting the contributor's repository and branch (use `gh pr create --repo <contributor-repo> --base <contributor-branch> --head diegosouzapw:fix-pr-<NUMBER>`).
-  4. Once they accept our PR into their branch, their original PR to our `main` will automatically update and become green.
-
-- Run the project's test suite locally to verify nothing breaks:
-  // turbo
-- Run: `npm test` or equivalent test command
-
-### 8. Merge into Release Branch
-
-### 8. Merge into Release Branch (NEVER CLOSE!)
-
-> **⚠️ CRITICAL**: NEVER use `gh pr close` for a PR whose idea or code was accepted. Closing a PR in a contributor's face after taking their idea—or closing it just because it had conflicts—is unacceptable.
-> You MUST ALWAYS resolve conflicts and apply fixes on the author's PR branch, and then merge the PR using GitHub so the contributor gets the official "Merged" badge and proper credit on their profile.
-
-Even if the PR had severe conflicts or required significant architectural adjustments, you MUST:
-
-1. Resolve any conflicts and apply the fixes directly to their PR branch (as detailed in step 7).
-2. Once the PR branch is green, conflict-free, and correct, merge it into the release branch using the GitHub CLI.
-
-```bash
-# Merge the PR (base is already set to release/vX.Y.Z from step 3.5)
-gh pr merge <NUMBER> --repo <owner>/<repo> --squash --body "Integrated into release/vX.Y.Z"
-```
-
-In ALL cases:
-
-- Post a **thank-you comment** on the PR via the GitHub API before or immediately after merging.
-- The message should:
-  - Thank the author by name/username for their contribution.
-  - Explain what was adjusted or improved (if we pushed fixes to their branch).
-  - Note it will be included in the upcoming release.
-  - Be friendly, professional, and encouraging.
-- Example: _"Thanks @author for this great contribution! 🎉 We've added a few small adjustments to your branch to align with our latest architecture, and it's now officially merged into the release/vX.Y.Z branch. It will be part of the next release. We appreciate your effort!"_
-
-### 9. Sync Local Release Branch
-
-After merging PRs, sync the local release branch to include the new changes:
-
-```bash
-git fetch origin
-git pull origin release/vX.Y.Z
-```
-
-### 10. Continue or Finalize
-
-After processing all approved PRs:
-
-- If more PRs remain, go back to step 7
-- When all PRs are processed, **update CHANGELOG.md** on the release branch with all new entries
-- Run **test coverage** to verify all metrics stay above 85%:
-  ```bash
-  npm run test:coverage
-  ```
-- Fix any test regressions introduced by merged PRs
-- Run `/generate-release` workflow Phase 1 steps 7–10 (tests → commit → push → open PR to main → wait for user)
-- The `/generate-release` workflow handles the final merge from `release/vX.Y.Z` → `main`

.agents/workflows/version-bump.md

@@ -1,327 +0,0 @@
----
-description: Bump version, auto-generate CHANGELOG from git commits, update all versioned files, and refresh root + docs/ documentation to reflect the current project state
----
-
-# Version Bump Workflow
-
-Automatically bump the project version, generate CHANGELOG entries from git history since the last tag, update every file that references the version, and refresh project documentation to reflect the current state.
-
-> **VERSION RULE: Always use PATCH bumps (3.x.y → 3.x.y+1)**
-> NEVER use `npm version minor` or `npm version major`.
-> Always use: `npm version patch --no-git-tag-version`
-> The threshold rule: when `y` reaches 10, bump to `3.(x+1).0` — e.g. `3.4.10` → `3.5.0`.
-
----
-
-## Phase 1: Determine Version
-
-### 1. Read current version and last tag
-
-// turbo
-
-```bash
-cd /home/diegosouzapw/dev/proxys/9router
-CURRENT_VERSION=$(node -p "require('./package.json').version")
-LAST_TAG=$(git describe --tags --abbrev=0 2>/dev/null || echo "")
-CURRENT_BRANCH=$(git branch --show-current)
-echo "Current version: $CURRENT_VERSION"
-echo "Last tag: $LAST_TAG"
-echo "Current branch: $CURRENT_BRANCH"
-```
-
-### 2. Calculate new version
-
-Apply the patch bump rule:
-
-- If the current patch number is `9`, the new version is `3.(minor+1).0`
-- Otherwise, increment patch: `3.x.y` → `3.x.(y+1)`
-
-If the version was ALREADY bumped (e.g. you are on a release branch and package.json already has the new version), **skip the npm version bump** and use the existing version.
-
-### 3. Bump package.json (if needed)
-
-// turbo
-
-```bash
-# Only if version hasn't been bumped yet
-npm version patch --no-git-tag-version
-```
-
-Or for threshold (y=10):
-
-```bash
-# Manual threshold bump
-VERSION="3.X.0"  # compute manually
-npm version "$VERSION" --no-git-tag-version
-```
-
----
-
-## Phase 2: Generate CHANGELOG from Git History
-
-### 4. Collect commits since last tag
-
-// turbo
-
-```bash
-cd /home/diegosouzapw/dev/proxys/9router
-LAST_TAG=$(git describe --tags --abbrev=0 2>/dev/null)
-echo "=== Commits since $LAST_TAG ==="
-git log "$LAST_TAG"..HEAD --pretty=format:"%h %s" --no-merges | head -100
-echo ""
-echo "=== Merge commits ==="
-git log "$LAST_TAG"..HEAD --merges --pretty=format:"%h %s" | head -50
-```
-
-### 5. Classify commits and generate CHANGELOG section
-
-Analyze each commit message and classify into categories based on the conventional-commit prefix and content:
-
-| Category            | Patterns                                         |
-| ------------------- | ------------------------------------------------ |
-| ✨ New Features     | `feat:`, `feat(*):`                              |
-| 🐛 Bug Fixes        | `fix:`, `fix(*):`                                |
-| ⚠️ Breaking Changes | `BREAKING CHANGE`, `!:` suffix                   |
-| 🛠️ Maintenance      | `chore:`, `refactor:`, `perf:`, `build:`         |
-| 🧪 Tests            | `test:`, `tests:`                                |
-| 📝 Documentation    | `docs:`                                          |
-| 🔒 Security         | `security:`, CVE references, vulnerability fixes |
-| 🌍 i18n             | translation updates, locale changes              |
-
-For each category with entries, create a markdown section with descriptive bullet points. Use the commit messages but rewrite them to be human-readable and descriptive (not raw commit messages).
-
-**If a commit references a PR number** (e.g. `#880`, `PR #885`), include it in the description.
-
-### 6. Update CHANGELOG.md
-
-Replace the `## [Unreleased]` section content with the generated entries, then add the new versioned section:
-
-```markdown
-## [Unreleased]
-
----
-
-## [NEW_VERSION] — YYYY-MM-DD
-
-### ✨ New Features
-
-- **Feature name:** Description (#PR)
-
-### 🐛 Bug Fixes
-
-- **Fix name:** Description (#PR)
-
-### 🛠️ Maintenance
-
-- **Item:** Description
-
----
-
-## [PREVIOUS_VERSION] — YYYY-MM-DD
-
-...
-```
-
-The date must be today's date in `YYYY-MM-DD` format.
-
----
-
-## Phase 3: Sync Version Across All Files
-
-### 7. Update workspace package.json files and openapi.yaml
-
-// turbo
-
-```bash
-cd /home/diegosouzapw/dev/proxys/9router
-VERSION=$(node -p "require('./package.json').version")
-
-# Update docs/openapi.yaml version
-sed -i "s/  version: .*/  version: $VERSION/" docs/openapi.yaml
-echo "✓ docs/openapi.yaml → $VERSION"
-
-# Update workspace packages (open-sse, electron)
-for dir in electron open-sse; do
-  if [ -d "$dir" ] && [ -f "$dir/package.json" ]; then
-    (cd "$dir" && npm version "$VERSION" --no-git-tag-version --allow-same-version > /dev/null)
-    echo "✓ $dir/package.json → $VERSION"
-  fi
-done
-
-echo "✓ All workspace packages synced to $VERSION"
-```
-
-### 8. Update llm.txt version references
-
-// turbo
-
-```bash
-cd /home/diegosouzapw/dev/proxys/9router
-VERSION=$(node -p "require('./package.json').version")
-OLD_VERSION_PATTERN='[0-9]\+\.[0-9]\+\.[0-9]\+'
-
-# Update "Current version:" line
-sed -i "s/\*\*Current version:\*\* $OLD_VERSION_PATTERN/**Current version:** $VERSION/" llm.txt
-
-# Update "Key Features (vX.Y.Z)" header
-sed -i "s/## Key Features (v$OLD_VERSION_PATTERN)/## Key Features (v$VERSION)/" llm.txt
-
-echo "✓ llm.txt → $VERSION"
-```
-
-### 9. Regenerate lock file
-
-// turbo
-
-```bash
-cd /home/diegosouzapw/dev/proxys/9router
-npm install
-echo "✓ Lock file regenerated"
-```
-
----
-
-## Phase 4: Update Root Documentation
-
-Based on the CHANGELOG entries generated in Phase 2, review and update these root-level files if relevant changes warrant updates:
-
-### 10. Review and update root documentation files
-
-For each file below, read the current content and determine if the CHANGELOG entries require any updates. Only modify files where substantive changes have occurred:
-
-| File              | When to update                                                                                                              |
-| ----------------- | --------------------------------------------------------------------------------------------------------------------------- |
-| `README.md`       | New providers, major features, stats changes (test count, provider count), badges, installation instructions, feature table |
-| `AGENTS.md`       | Architecture changes, new modules, new commands, new providers, new services/handlers/executors                             |
-| `CONTRIBUTING.md` | Dev workflow changes, new tooling, test infrastructure changes                                                              |
-| `SECURITY.md`     | Security fixes, new auth mechanisms, vulnerability disclosures                                                              |
-| `llm.txt`         | Provider count changes, new features, architecture changes                                                                  |
-
-**Update rules:**
-
-- **README.md**: Update provider count, test count, feature highlights table, badges if any numbers changed. If a new provider was added, add it to the provider table. If a major feature was added, add it to the features section.
-- **AGENTS.md**: If new architecture components (handlers, executors, services, DB modules) were added, update the Architecture section. If new commands were added, update the Build/Test table.
-- **SECURITY.md**: Add new vulnerability fixes or security improvements to the relevant section.
-- **llm.txt**: Update provider count, feature list, version references.
-
-### 11. Review and update docs/ files (excluding i18n/)
-
-For each file in `docs/` (excluding `docs/i18n/`), review if CHANGELOG changes affect it:
-
-| File                             | When to update                                      |
-| -------------------------------- | --------------------------------------------------- |
-| `docs/API_REFERENCE.md`          | New API endpoints, changed request/response formats |
-| `docs/ARCHITECTURE.md`           | New modules, new services, changed data flow        |
-| `docs/CLI-TOOLS.md`              | New CLI tool integrations, config format changes    |
-| `docs/FEATURES.md`               | New features, removed features, changed settings    |
-| `docs/MCP-SERVER.md`             | New MCP tools, changed tool signatures              |
-| `docs/A2A-SERVER.md`             | New A2A skills, protocol changes                    |
-| `docs/USER_GUIDE.md`             | UX changes, new dashboard pages, settings changes   |
-| `docs/VM_DEPLOYMENT_GUIDE.md`    | Deployment changes, new env vars                    |
-| `docs/TROUBLESHOOTING.md`        | New known issues, resolved problems                 |
-| `docs/AUTO-COMBO.md`             | Routing changes, new strategies                     |
-| `docs/CODEBASE_DOCUMENTATION.md` | New files, architectural changes                    |
-| `docs/RELEASE_CHECKLIST.md`      | Process changes                                     |
-| `docs/COVERAGE_PLAN.md`          | Test changes                                        |
-| `docs/openapi.yaml`              | Already updated in step 7                           |
-
-**Only update files where the CHANGELOG entries directly affect the documented content.** Do NOT update files just to bump a version number — only when the documented behavior, features, or architecture has actually changed.
-
----
-
-## Phase 5: Verify
-
-### 12. Run lint check
-
-// turbo
-
-```bash
-cd /home/diegosouzapw/dev/proxys/9router
-npm run lint
-```
-
-### 13. Run tests
-
-// turbo
-
-```bash
-cd /home/diegosouzapw/dev/proxys/9router
-npm test
-```
-
-### 14. Verify version sync across all files
-
-// turbo
-
-```bash
-cd /home/diegosouzapw/dev/proxys/9router
-VERSION=$(node -p "require('./package.json').version")
-echo "Expected version: $VERSION"
-echo ""
-
-echo "--- package.json ---"
-grep '"version"' package.json | head -1
-
-echo "--- open-sse/package.json ---"
-grep '"version"' open-sse/package.json | head -1
-
-echo "--- electron/package.json ---"
-[ -f electron/package.json ] && grep '"version"' electron/package.json | head -1
-
-echo "--- docs/openapi.yaml ---"
-grep "  version:" docs/openapi.yaml | head -1
-
-echo "--- llm.txt ---"
-grep "Current version:" llm.txt
-
-echo "--- CHANGELOG.md (first versioned entry) ---"
-grep "^## \[" CHANGELOG.md | head -2
-```
-
-### 15. 🛑 STOP — Present Summary to User
-
-**STOP** and present a summary to the user including:
-
-- Old version → New version
-- CHANGELOG entries generated
-- Files modified
-- Test results
-- Any documentation updates made
-
-**Wait for the user to confirm before committing.**
-
----
-
-## Phase 6: Commit (only after user approval)
-
-### 16. Stage and commit
-
-// turbo-all
-
-```bash
-cd /home/diegosouzapw/dev/proxys/9router
-git add -A
-VERSION=$(node -p "require('./package.json').version")
-git commit -m "chore(release): bump to v$VERSION — changelog, docs, version sync"
-```
-
----
-
-## Notes
-
-- This workflow does **NOT** create tags, releases, or deploy. Use `/generate-release` for the full release cycle after this.
-- This workflow does **NOT** update `docs/i18n/` translations. Use `/update-i18n` separately after committing.
-- The CHANGELOG generation is based on git commits since the last tag. If there are no new commits, the workflow should inform the user and stop.
-- Always verify the generated CHANGELOG entries make sense — raw commit messages may need rewriting for clarity.
-- If the version was already bumped (e.g. you're on a `release/vX.Y.Z` branch), skip the `npm version` step and use the existing version.
-
-## Version Touchpoints Checklist
-
-| File                    | Field/Pattern                                               |
-| ----------------------- | ----------------------------------------------------------- |
-| `package.json`          | `"version": "X.Y.Z"`                                        |
-| `open-sse/package.json` | `"version": "X.Y.Z"`                                        |
-| `electron/package.json` | `"version": "X.Y.Z"`                                        |
-| `docs/openapi.yaml`     | `version: X.Y.Z`                                            |
-| `llm.txt`               | `**Current version:** X.Y.Z` and `## Key Features (vX.Y.Z)` |
-| `CHANGELOG.md`          | `## [X.Y.Z] — YYYY-MM-DD`                                   |

.dockerignore

@@ -9,6 +9,7 @@
 # Dependencies and build output
 node_modules
 .next
+.build
 out
 build
 dist
@@ -37,8 +38,24 @@ test-results
 playwright-report
 blob-report
 
-# Documentation (not needed in container)
-docs
+# 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
 
@@ -60,6 +77,8 @@ bun.lock
 # Agent config
 .agents
 .gemini
+.claude
+.source
 
 # Misc
 llm.txt

.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

@@ -13,7 +13,7 @@ body:
     attributes:
       label: OmniRoute Version
       description: "Run `omniroute --version` or check the left sidebar in the dashboard."
-      placeholder: "e.g. 3.0.9"
+      placeholder: "e.g. 3.7.9"
     validations:
       required: true
 
@@ -44,7 +44,7 @@ body:
     id: os-version
     attributes:
       label: OS Version
-      placeholder: "e.g. Windows 11 23H2, macOS 15.3, Ubuntu 24.04"
+      placeholder: "e.g. Windows 11 25H2, macOS 26.5, Ubuntu 26.04"
     validations:
       required: false
 
@@ -53,7 +53,7 @@ body:
     attributes:
       label: Node.js Version
       description: "Run `node --version`. Skip if using Docker."
-      placeholder: "e.g. 22.12.0"
+      placeholder: "e.g. 24.15.0"
     validations:
       required: false
 
@@ -70,7 +70,7 @@ body:
     id: model
     attributes:
       label: Model(s) Involved
-      placeholder: "e.g. claude-sonnet-4-20250514, gpt-4o, gemini-2.5-pro"
+      placeholder: "e.g. claude-opus-4-7, gpt-5.5, gemini-3.1-pro"
     validations:
       required: false
 
@@ -165,7 +165,7 @@ body:
       description: "Which commands or tests should prove this bug is fixed?"
       placeholder: |
         Example:
-        - node --import tsx/esm --test tests/unit/my-file.test.ts
+        - node --import tsx --test tests/unit/my-file.test.ts
         - npm run test:coverage
     validations:
       required: false

.github/ISSUE_TEMPLATE/test_coverage_task.yml

@@ -59,7 +59,7 @@ body:
       description: "List the commands that must pass before this issue can be closed."
       placeholder: |
         Example:
-        - node --import tsx/esm --test tests/unit/my-suite.test.ts
+        - node --import tsx --test tests/unit/my-suite.test.ts
         - npm run test:coverage
     validations:
       required: true

.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/dependabot.yml

@@ -24,6 +24,20 @@ 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: "/"

.github/pull_request_template.md

@@ -27,4 +27,4 @@
 
 ## Reviewer Notes
 
-- Call out any risky areas, migrations, feature flags, or manual validation that reviewers should know about.
+- Call out any risky areas, migrations, feature flags, or manual validation that reviewers should know about.

.github/workflows/build-fork.yml

@@ -1,21 +1,36 @@
-name: Build Fork Image (ghcr.io)
+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
-  packages: write
+
+env:
+  IMAGE_NAME: ghcr.io/kang-heewon/omniroute
 
 jobs:
   build:
-    name: Build and Push to ghcr.io
+    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@v6
+        uses: actions/checkout@v7
         with:
-          ref: fix/claude-thinking-mapping
+          persist-credentials: false
+
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@v4
 
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v4
@@ -27,15 +42,30 @@ jobs:
           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
+          platforms: linux/amd64,linux/arm64
           push: true
-          tags: |
-            ghcr.io/inkorcloud/omniroute:fix-claude-thinking-mapping
-            ghcr.io/inkorcloud/omniroute:latest
+          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/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

@@ -17,27 +17,82 @@ jobs:
     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
-        continue-on-error: true
         with:
           host: ${{ secrets.VPS_HOST }}
           username: ${{ secrets.VPS_USER }}
           key: ${{ secrets.VPS_SSH_KEY }}
           port: 22
-          timeout: 30s
-          command_timeout: 5m
+          timeout: 60s
+          command_timeout: 15m
           script: |
-            echo "=== Updating OmniRoute ==="
-            npm install -g omniroute@latest 2>&1
-            INSTALLED_VERSION=$(omniroute --version 2>/dev/null || echo "unknown")
-            echo "Installed version: $INSTALLED_VERSION"
+            set -euo pipefail
 
-            echo "=== Restarting PM2 ==="
-            pm2 restart omniroute || pm2 start omniroute --name omniroute -- --port 20128
+            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
 
-            echo "=== Health Check ==="
-            sleep 3
-            curl -sf http://localhost:20128/api/settings > /dev/null && echo "✅ OmniRoute is healthy" || echo "❌ Health check failed"
+            # 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

@@ -4,33 +4,150 @@ 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. 2.6.0)"
+        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
-  packages: write
 
 jobs:
-  docker:
-    name: Build and Push Docker (multi-arch)
+  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: Set up QEMU (for multi-arch builds)
-        uses: docker/setup-qemu-action@v4
+      - 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: ${{ inputs.promote_latest }}
+        run: |
+          set -euo pipefail
+
+          # 1) Resolve version string from the trigger (all inputs come via env).
+          case "$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" >> "$GITHUB_OUTPUT"
+
+          # 2) Decide whether to promote :latest.
+          PROMOTE="false"
+          if [ "$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@v4
@@ -48,41 +165,227 @@ jobs:
           username: ${{ github.actor }}
           password: ${{ secrets.GITHUB_TOKEN }}
 
-      - name: Extract version from release tag or input
-        id: version
-        run: |
-          if [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
-            VERSION="${{ inputs.version }}"
-          else
-            VERSION="${GITHUB_REF_NAME}"
-            VERSION="${VERSION#v}"
-          fi
-          echo "version=$VERSION" >> "$GITHUB_OUTPUT"
-          echo "Publishing Docker image: $IMAGE_NAME:$VERSION"
-
-      - name: Build and push multi-arch image
+      - name: Build and push platform image by digest
+        id: build
         uses: docker/build-push-action@v7
         with:
           context: .
           target: runner-base
-          platforms: linux/amd64,linux/arm64
-          push: true
+          platforms: ${{ matrix.platform }}
+          outputs: type=image,push-by-digest=true,name-canonical=true,push=true
           tags: |
-            ${{ env.IMAGE_NAME }}:${{ steps.version.outputs.version }}
-            ${{ env.IMAGE_NAME }}:latest
-            ghcr.io/diegosouzapw/omniroute:${{ steps.version.outputs.version }}
-            ghcr.io/diegosouzapw/omniroute:latest
-          cache-from: type=gha
-          cache-to: type=gha,mode=max
+            ${{ 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: Inspect image
+      - 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: ${{ steps.build-web.outputs.digest }}
         run: |
-          docker buildx imagetools inspect "${{ env.IMAGE_NAME }}:${{ steps.version.outputs.version }}"
+          set -euo pipefail
+          mkdir -p /tmp/digests/base /tmp/digests/web
+          touch "/tmp/digests/base/${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

@@ -11,43 +11,56 @@ on:
         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: write
-  id-token: write
-  packages: write
+  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@v6
+        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: ${{ inputs.version }}
         run: |
-          if [[ "${{ github.event_name }}" == "push" ]]; then
+          if [[ "$EVENT_NAME" == "push" ]]; then
             VERSION="${GITHUB_REF#refs/tags/}"
           else
-            VERSION="${{ inputs.version }}"
+            VERSION="$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" >> $GITHUB_OUTPUT
+          echo "version=$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:
@@ -71,7 +84,9 @@ jobs:
             deb_ext: .deb
 
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
       - name: Setup Node
         uses: actions/setup-node@v6
         with:
@@ -79,7 +94,7 @@ jobs:
           cache: npm
 
       - name: Cache node_modules
-        uses: actions/cache@v5
+        uses: actions/cache@v6.0.0
         with:
           path: node_modules
           key: ${{ runner.os }}-node-${{ hashFiles('package-lock.json') }}
@@ -99,17 +114,23 @@ jobs:
           # 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" >> $GITHUB_ENV
+          echo "USERPROFILE=$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: ${{ needs.validate.outputs.version }}
         run: |
-          VERSION="${{ needs.validate.outputs.version }}"
           VERSION_NO_V="${VERSION#v}"
           node -e "
             const fs = require('fs');
@@ -135,9 +156,16 @@ jobs:
 
       - name: Smoke packaged Electron app
         if: matrix.platform != 'linux'
-        # Windows CI: requestSingleInstanceLock() fails due to USERPROFILE
-        # sanitization needed for the build step. Smoke is best-effort there.
-        continue-on-error: ${{ matrix.platform == 'windows' }}
+        # 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"
@@ -184,10 +212,13 @@ jobs:
     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@v6
+        uses: actions/checkout@v7
         with:
+          persist-credentials: false
           fetch-depth: 0
 
       - name: Download all artifacts
@@ -197,14 +228,20 @@ jobs:
           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: ${{ needs.validate.outputs.version }}
         run: |
           # Create source code archives (excluding dev dependencies and build artifacts)
-          export TARBALL="OmniRoute-${{ needs.validate.outputs.version }}.source.tar.gz"
-          export ZIPBALL="OmniRoute-${{ needs.validate.outputs.version }}.source.zip"
+          export TARBALL="OmniRoute-${VERSION}.source.tar.gz"
+          export ZIPBALL="OmniRoute-${VERSION}.source.zip"
 
           # Use git archive for clean source export
-          git archive --format=tar.gz --prefix=OmniRoute-${{ needs.validate.outputs.version }}/ HEAD -o "release-assets/$TARBALL"
-          git archive --format=zip --prefix=OmniRoute-${{ needs.validate.outputs.version }}/ HEAD -o "release-assets/$ZIPBALL"
+          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"
@@ -234,6 +271,14 @@ jobs:
   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 }}

.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: ${{ matrix.batch.timeout || 180 }}
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: "24"
+          cache: npm
+      - run: npm ci
+      - name: Restore Stryker incremental cache
+        uses: actions/cache@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        with:
+          path: reports/mutation/stryker-incremental.json
+          key: stryker-incremental-${{ 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@v6.0.0
+        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

@@ -1,8 +1,11 @@
 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:
@@ -10,13 +13,15 @@ on:
         required: true
         type: string
       tag:
-        description: "npm dist-tag (latest / next)"
+        description: "npm dist-tag (auto / latest / next / historic)"
         required: false
-        default: "latest"
+        default: "auto"
         type: choice
         options:
+          - auto
           - latest
           - next
+          - historic
   workflow_call:
     inputs:
       version:
@@ -24,18 +29,19 @@ on:
         required: true
         type: string
       tag:
-        description: "npm dist-tag (latest / next)"
+        description: "npm dist-tag (auto / latest / next / historic)"
         required: false
-        default: "latest"
+        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
-  packages: write
 
 env:
   NPM_PUBLISH_NODE_VERSION: "24"
@@ -43,10 +49,18 @@ env:
 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
@@ -57,77 +71,172 @@ jobs:
       - name: Install dependencies (skip scripts to avoid heavy build)
         run: npm install --ignore-scripts --no-audit --no-fund
 
-      - name: Resolve version and dist-tag
+      - 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: ${{ inputs.tag }}
         run: |
-          VERSION="${{ inputs.version }}"
-          TAG="${{ inputs.tag }}"
+          set -euo pipefail
 
-          if [ -z "$VERSION" ]; then
-            if [ "${{ github.event_name }}" = "release" ]; then
-              VERSION="${GITHUB_REF_NAME}"
-            fi
+          # 1) Resolve VERSION from the trigger (all inputs come via env).
+          VERSION="${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
 
-          # Strip v prefix if present
-          VERSION="${VERSION#v}"
-
-          # Default dist-tag logic
-          if [ -z "$TAG" ]; then
-            if [[ "$VERSION" == *-* ]]; then
+          # 2) Resolve dist-tag.
+          #    - explicit 'latest'/'next'/'historic' is honored
+          #    - 'auto' (or empty): pre-release identifiers → 'next';
+          #      stable versions → 'latest' only if VERSION is the highest
+          #      stable semver among `v*` tags (otherwise → 'historic').
+          REQUESTED_TAG="${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
-              TAG="latest"
+              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
-          echo "version=$VERSION" >> $GITHUB_OUTPUT
-          echo "tag=$TAG" >> $GITHUB_OUTPUT
-          echo "📦 Publishing omniroute@$VERSION with tag=$TAG"
+
+          # 3) Skip-if-already-published. NOTE: do NOT pass `--silent` to
+          #    `npm view` — it suppresses stdout and breaks the grep, which
+          #    caused old releases (3.2.8) to be re-published and steal
+          #    dist-tag `latest`. See incident notes in CHANGELOG.
+          PUBLISHED="$(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: ${{ steps.resolve.outputs.version }}
         run: |
-          npm version "${{ steps.resolve.outputs.version }}" --no-git-tag-version --allow-same-version
+          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: Validate npm package artifact
+        if: steps.resolve.outputs.skip != 'true'
         run: npm run check:pack-artifact
 
-      - name: Publish to npm
-        run: |
-          VERSION="${{ steps.resolve.outputs.version }}"
-          TAG="${{ steps.resolve.outputs.tag }}"
-          # Check if this version is already published — skip instead of failing with E403
-          if npm view "omniroute@${VERSION}" version --silent 2>/dev/null | grep -q "^${VERSION}$"; then
-            echo "⚠️ Version ${VERSION} is already published on npm — skipping."
-            exit 0
-          fi
-          if [ "$TAG" = "latest" ]; then
-            npm publish --access public
-          else
-            npm publish --access public --tag "$TAG"
-          fi
-          echo "✅ Published omniroute@$VERSION (tag: $TAG)"
-        env:
-          NODE_AUTH_TOKEN: ${{ secrets.NPM_TOKEN }}
+      - 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: Publish to GitHub Packages
-        run: |
-          VERSION="${{ steps.resolve.outputs.version }}"
-          TAG="${{ steps.resolve.outputs.tag }}"
+      - 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
 
-          echo "Configuring for GitHub Packages..."
-          echo "//npm.pkg.github.com/:_authToken=${{ secrets.GITHUB_TOKEN }}" > .npmrc
-          npm pkg set name="@diegosouzapw/omniroute"
-
-          if [ "$TAG" = "latest" ]; then
-            npm publish --registry=https://npm.pkg.github.com || echo "⚠️ Version ${VERSION} might already be published on GitHub."
-          else
-            npm publish --registry=https://npm.pkg.github.com --tag "$TAG" || echo "⚠️ Version ${VERSION} might already be published on GitHub."
-          fi
-          echo "✅ Action finished for GitHub Packages"
+      - 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
+        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 +1,144 @@
+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
+      - name: Docs accuracy (fabricated-docs + i18n mirrors, strict)
+        run: npm run check:docs-all
+      - 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:test-runner-api
+      - run: npm run check:any-budget:t11
+      # Build-scope guard: fails if worktrees/cruft leak into the tsconfig include
+      # scope (would OOM `next build`). Instant. See incident 2026-06-25 / #5031.
+      - run: npm run check:build-scope
+      # Pack-policy (unexpected-files allowlist) WITHOUT a build — catches a stray file
+      # leaking into the npm tarball (v3.8.36: 6 ops bin/*.sh) per-PR instead of only on
+      # the release PR's heavy Package Artifact job.
+      - run: npm run check:pack-policy
+      # Complexity + cognitive-complexity ratchets on the fast-path (PR→release) so
+      # cycle drift is rebaselined PER-PR instead of cascading onto the release PR's
+      # Quality Ratchet (v3.8.36: +30 complexity / +15 cognitive surfaced only post-merge).
+      - run: npm run check:complexity
+      - run: npm run check:cognitive-complexity
+      - 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[@]}"
+
+  fast-vitest:
+    name: Vitest (fast-path)
+    runs-on: ubuntu-latest
+    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:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: ${{ env.CI_NODE_VERSION }}
+          cache: npm
+      - run: npm ci
+      - run: npm run test:vitest
+
+  fast-unit:
+    name: Unit Tests fast-path (${{ matrix.shard }}/2)
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        shard: [1, 2]
+    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:
+          persist-credentials: false
+      - uses: actions/setup-node@v6
+        with:
+          node-version: ${{ env.CI_NODE_VERSION }}
+          cache: npm
+      - run: npm ci
+      - run: >
+          node --max-old-space-size=4096 --import tsx
+          --import ./tests/_setup/isolateDataDir.ts
+          --test --test-force-exit --test-concurrency=4 --test-shard=${{ matrix.shard }}/2
+          tests/unit/*.test.ts
+          "tests/unit/{api,auth,authz,build,cli,cli-helper,combo,compression,correctness,cors,dashboard,db,db-adapters,docs,gamification,guardrails,lib,mcp,runtime,security,services,settings,shared,ui}/**/*.test.ts"

.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

@@ -4,6 +4,42 @@
 .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)
 /_*/
@@ -13,32 +49,19 @@ docs/new-features/
 
 # dependencies
 node_modules/
-/.pnp
-.pnp.*
-.yarn/*
-!.yarn/patches
-!.yarn/plugins
-!.yarn/releases
-!.yarn/versions
-.data/
-.next-playwright/
-
-# testing
-coverage/
-coverage**
-
-# next.js
-.next/
-/out/
-
-# production
-/build
-/app
-cloud/*
-
-# misc
+# 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
+*.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*
@@ -49,6 +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
@@ -61,6 +87,7 @@ next-env.d.ts
 data/
 .data/
 logs/*
+test_output.log
 
 # analysis directories (generated, not tracked)
 .analysis/
@@ -68,45 +95,6 @@ antigravity-manager-analysis/
 .sisyphus/
 .plans/
 
-# 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/RELEASE_CHECKLIST.md
-!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/
-!docs/i18n/
-!docs/i18n/**
-!docs/features/
-!docs/features/**
-!docs/A2A-SERVER.md
-!docs/AUTO-COMBO.md
-!docs/MCP-SERVER.md
-!docs/CLI-TOOLS.md
-!docs/COVERAGE_PLAN.md
-!docs/ENVIRONMENT.md
-!docs/UNINSTALL.md
-!docs/I18N.md
-!docs/FLY_IO_DEPLOYMENT_GUIDE.md
-
-
 # open-sse tests
 open-sse/test/*
 
@@ -114,6 +102,7 @@ open-sse/test/*
 .github/instructions/codacy.instructions.md
 
 # Playwright
+.playwright-mcp/
 test-results/
 playwright-report/
 blob-report/
@@ -128,18 +117,24 @@ clipr/
 app.log
 *.tgz
 .gh-discussions.json
+deploy.sh
+docker-compose.minimal.yml
+
 
 # Backup directories
 app.__qa_backup/
 .app-build-backup-*/
+backup/
 
-# Production standalone build (created by scripts/prepublish.mjs)
-# Conflicts with Next.js App Router detection in dev (root app/ shadows src/app/)
-# npm publish still includes it via package.json "files" field
-/app/
+# 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 (subproject dependency lock and build artifacts)
-electron/package-lock.json
+# Electron
 electron/dist-electron/
 electron/node_modules/
 icon.iconset/
@@ -152,9 +147,6 @@ vscode-extension/
 *.sqlite-wal
 *.sqlite-journal
 
-# Compiled npm-package build artifact (not source, should not be in git)
-/app
-
 # IDEA
 .idea/
 
@@ -169,6 +161,11 @@ typescript
 
 # Superpowers plans/specs (internal tooling, not project code)
 docs/superpowers/
+# Superpowers visual-companion brainstorm mockups (ephemeral)
+.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
@@ -182,3 +179,53 @@ 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

@@ -5,6 +5,9 @@ if ! command -v npx >/dev/null 2>&1; then
   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-docs-sync.mjs
+node scripts/check/check-docs-sync.mjs
 npm run check:any-budget:t11
+node scripts/check/check-tracked-artifacts.mjs

.husky/pre-push

@@ -1,8 +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 test' manually before pushing."
+  echo "   Run 'npm run check:any-budget:t11 && npm run check:tracked-artifacts' manually before pushing."
   exit 0
 fi
 
-npm run test:unit
+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."
+    }
+  }
+}

.npmignore

@@ -9,6 +9,16 @@ app/vscode-extension/
 **/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/
@@ -18,6 +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
@@ -52,8 +73,8 @@ 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
@@ -100,3 +121,4 @@ test-results/
 playwright-report/
 blob-report/
 coverage/
+@omniroute/

.omc/project-memory.json

@@ -1,250 +0,0 @@
-{
-  "version": "1.0.0",
-  "lastScanned": 1775016362438,
-  "projectRoot": "/home/openclaw/omniroute-src",
-  "techStack": {
-    "languages": [
-      {
-        "name": "JavaScript/TypeScript",
-        "version": ">=18.0.0 <24.0.0",
-        "confidence": "high",
-        "markers": ["package.json"]
-      },
-      {
-        "name": "TypeScript",
-        "version": null,
-        "confidence": "high",
-        "markers": ["tsconfig.json"]
-      }
-    ],
-    "frameworks": [
-      {
-        "name": "express",
-        "version": "5.2.1",
-        "category": "backend"
-      },
-      {
-        "name": "next",
-        "version": "16.0.10",
-        "category": "fullstack"
-      },
-      {
-        "name": "react",
-        "version": "19.2.4",
-        "category": "frontend"
-      },
-      {
-        "name": "react-dom",
-        "version": "19.2.4",
-        "category": "frontend"
-      },
-      {
-        "name": "@playwright/test",
-        "version": "1.58.2",
-        "category": "testing"
-      },
-      {
-        "name": "vitest",
-        "version": "4.0.18",
-        "category": "testing"
-      }
-    ],
-    "packageManager": "npm",
-    "runtime": "Node.js 18.0.024.0.0"
-  },
-  "build": {
-    "buildCommand": "npm run build",
-    "testCommand": "npm test",
-    "lintCommand": "npm run lint",
-    "devCommand": "npm run dev",
-    "scripts": {
-      "dev": "node scripts/run-next.mjs dev",
-      "build": "node scripts/build-next-isolated.mjs",
-      "build:cli": "node scripts/prepublish.mjs",
-      "start": "node scripts/run-next.mjs start",
-      "lint": "eslint .",
-      "electron:dev": "concurrently \"npm run dev\" \"wait-on http://localhost:20128 && cd electron && npm run dev\"",
-      "electron:build": "npm run build && cd electron && npm run build",
-      "electron:build:win": "npm run build && cd electron && npm run build:win",
-      "electron:build:mac": "npm run build && cd electron && npm run build:mac",
-      "electron:build:linux": "npm run build && cd electron && npm run build:linux",
-      "test": "node --import tsx/esm --test tests/unit/*.test.mjs",
-      "test:unit": "node --import tsx/esm --test tests/unit/*.test.mjs",
-      "test:plan3": "node --import tsx/esm --test tests/unit/plan3-p0.test.mjs",
-      "test:fixes": "node --import tsx/esm --test tests/unit/fixes-p1.test.mjs",
-      "test:security": "node --import tsx/esm --test tests/unit/security-fase01.test.mjs",
-      "check:cycles": "node scripts/check-cycles.mjs",
-      "check:route-validation:t06": "node scripts/check-route-validation.mjs",
-      "check:any-budget:t11": "node scripts/check-t11-any-budget.mjs",
-      "check:docs-sync": "node scripts/check-docs-sync.mjs",
-      "typecheck:core": "tsc --pretty false -p tsconfig.typecheck-core.json",
-      "typecheck:noimplicit:core": "tsc --pretty false -p tsconfig.typecheck-noimplicit-core.json",
-      "test:integration": "node --import tsx/esm --test tests/integration/*.test.mjs",
-      "test:e2e": "node scripts/run-playwright-tests.mjs test tests/e2e/*.spec.ts",
-      "test:protocols:e2e": "node scripts/run-protocol-clients-tests.mjs",
-      "test:vitest": "vitest run open-sse/mcp-server/__tests__/*.test.ts open-sse/services/autoCombo/__tests__/*.test.ts",
-      "test:ecosystem": "node scripts/run-ecosystem-tests.mjs",
-      "test:coverage": "c8 --exclude=tests/** --exclude=**/*.test.* --reporter=text-summary --reporter=html --reporter=json-summary --reporter=lcov --check-coverage --statements 55 --lines 55 --functions 55 --branches 60 node --import tsx/esm --test tests/unit/*.test.mjs",
-      "test:coverage:legacy": "c8 --exclude=open-sse --check-coverage --lines 50 --functions 50 --branches 50 node --import tsx/esm --test tests/unit/*.test.mjs",
-      "coverage:report": "c8 report --exclude=tests/** --exclude=**/*.test.* --reporter=text --reporter=text-summary --reporter=html --reporter=json-summary --reporter=lcov",
-      "coverage:report:legacy": "c8 report --exclude=open-sse --reporter=text --reporter=text-summary",
-      "test:all": "npm run test:unit && npm run test:vitest && npm run test:ecosystem && npm run test:e2e",
-      "check": "npm run lint && npm run test",
-      "prepublishOnly": "npm run build:cli",
-      "postinstall": "node scripts/postinstall.mjs",
-      "prepare": "husky",
-      "system-info": "node scripts/system-info.mjs"
-    }
-  },
-  "conventions": {
-    "namingStyle": "camelCase",
-    "importStyle": null,
-    "testPattern": null,
-    "fileOrganization": null
-  },
-  "structure": {
-    "isMonorepo": true,
-    "workspaces": ["open-sse"],
-    "mainDirectories": ["bin", "docs", "public", "scripts", "src", "tests"],
-    "gitBranches": {
-      "defaultBranch": "main",
-      "branchingStrategy": null
-    }
-  },
-  "customNotes": [],
-  "directoryMap": {
-    "bin": {
-      "path": "bin",
-      "purpose": "Executable scripts",
-      "fileCount": 3,
-      "lastAccessed": 1775016362426,
-      "keyFiles": ["mcp-server.mjs", "omniroute.mjs", "reset-password.mjs"]
-    },
-    "docs": {
-      "path": "docs",
-      "purpose": "Documentation",
-      "fileCount": 14,
-      "lastAccessed": 1775016362426,
-      "keyFiles": [
-        "A2A-SERVER.md",
-        "API_REFERENCE.md",
-        "ARCHITECTURE.md",
-        "AUTO-COMBO.md",
-        "CLI-TOOLS.md"
-      ]
-    },
-    "electron": {
-      "path": "electron",
-      "purpose": null,
-      "fileCount": 5,
-      "lastAccessed": 1775016362431,
-      "keyFiles": ["README.md", "main.js", "package.json", "preload.js", "types.d.ts"]
-    },
-    "images": {
-      "path": "images",
-      "purpose": null,
-      "fileCount": 1,
-      "lastAccessed": 1775016362434,
-      "keyFiles": ["omniroute.png"]
-    },
-    "logs": {
-      "path": "logs",
-      "purpose": null,
-      "fileCount": 3,
-      "lastAccessed": 1775016362434,
-      "keyFiles": ["build_clean_tools.log", "build_debug.log", "build_force_clean.log"]
-    },
-    "open-sse": {
-      "path": "open-sse",
-      "purpose": null,
-      "fileCount": 5,
-      "lastAccessed": 1775016362434,
-      "keyFiles": ["index.ts", "package.json", "tsconfig.json", "types.d.ts"]
-    },
-    "public": {
-      "path": "public",
-      "purpose": "Public files",
-      "fileCount": 3,
-      "lastAccessed": 1775016362435,
-      "keyFiles": ["apple-touch-icon.svg", "favicon.svg", "icon-192.svg"]
-    },
-    "scripts": {
-      "path": "scripts",
-      "purpose": "Build/utility scripts",
-      "fileCount": 23,
-      "lastAccessed": 1775016362435,
-      "keyFiles": [
-        "bootstrap-env.mjs",
-        "build-next-isolated.mjs",
-        "check-cycles.mjs",
-        "check-docs-sync.mjs",
-        "check-route-validation.mjs"
-      ]
-    },
-    "src": {
-      "path": "src",
-      "purpose": "Source code",
-      "fileCount": 4,
-      "lastAccessed": 1775016362435,
-      "keyFiles": ["instrumentation-node.ts", "instrumentation.ts", "proxy.ts", "server-init.ts"]
-    },
-    "tests": {
-      "path": "tests",
-      "purpose": "Test files",
-      "fileCount": 0,
-      "lastAccessed": 1775016362435,
-      "keyFiles": []
-    },
-    "electron/assets": {
-      "path": "electron/assets",
-      "purpose": "Static assets",
-      "fileCount": 4,
-      "lastAccessed": 1775016362436,
-      "keyFiles": ["icon.icns", "icon.ico", "icon.png"]
-    },
-    "open-sse/config": {
-      "path": "open-sse/config",
-      "purpose": "Configuration files",
-      "fileCount": 17,
-      "lastAccessed": 1775016362436,
-      "keyFiles": ["audioRegistry.ts", "cliFingerprints.ts", "codexInstructions.ts"]
-    },
-    "open-sse/services": {
-      "path": "open-sse/services",
-      "purpose": "Business logic services",
-      "fileCount": 35,
-      "lastAccessed": 1775016362437,
-      "keyFiles": ["accountFallback.ts", "accountSelector.ts", "apiKeyRotator.ts"]
-    },
-    "src/app": {
-      "path": "src/app",
-      "purpose": "Application code",
-      "fileCount": 7,
-      "lastAccessed": 1775016362438,
-      "keyFiles": ["error.tsx", "global-error.tsx", "globals.css"]
-    },
-    "src/lib": {
-      "path": "src/lib",
-      "purpose": "Library code",
-      "fileCount": 30,
-      "lastAccessed": 1775016362438,
-      "keyFiles": ["apiBridgeServer.ts", "apiKeyExposure.ts", "cacheControlSettings.ts"]
-    },
-    "src/middleware": {
-      "path": "src/middleware",
-      "purpose": "Middleware",
-      "fileCount": 1,
-      "lastAccessed": 1775016362438,
-      "keyFiles": ["promptInjectionGuard.ts"]
-    },
-    "src/models": {
-      "path": "src/models",
-      "purpose": "Data models",
-      "fileCount": 1,
-      "lastAccessed": 1775016362438,
-      "keyFiles": ["index.ts"]
-    }
-  },
-  "hotPaths": [],
-  "userDirectives": []
-}

.omc/sessions/53c002c3-36a6-47c3-a52d-a8f756c264eb.json

@@ -1,8 +0,0 @@
-{
-  "session_id": "53c002c3-36a6-47c3-a52d-a8f756c264eb",
-  "ended_at": "2026-04-01T04:06:04.924Z",
-  "reason": "prompt_input_exit",
-  "agents_spawned": 0,
-  "agents_completed": 0,
-  "modes_used": []
-}

.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": {
@@ -17,5 +18,117 @@
       "level": "off"
     }
   },
-  "git.ignoreLimitWarning": true
+  "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

@@ -3,19 +3,53 @@
 ## Project
 
 Unified AI proxy/router — route any LLM through one endpoint. Multi-provider support
-with **160+ providers** (OpenAI, Anthropic, Gemini, DeepSeek, Groq, xAI, Mistral, Fireworks,
+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** (29 tools), **A2A v0.3 Protocol**, and **Electron desktop app**.
+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 ≥18 <24, ES Modules (`"type": "module"`)
-- **Language**: TypeScript 5.9 (`src/`) + JavaScript (`open-sse/`, `electron/`)
+- **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 workspace package
 - **Styling**: Tailwind CSS v4
-- **i18n**: next-intl with 40+ languages
+- **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
 
@@ -23,19 +57,32 @@ with **MCP Server** (29 tools), **A2A v0.3 Protocol**, and **Electron desktop ap
 
 ## Build, Lint, and Test Commands
 
-| Command                             | Description                       |
-| ----------------------------------- | --------------------------------- |
-| `npm run dev`                       | Start Next.js dev server          |
-| `npm run build`                     | Production build (isolated)       |
-| `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 |
+| 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
 
@@ -120,6 +167,10 @@ Always run `prettier --write` on changed files.
 - 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.).
 
 ---
 
@@ -127,23 +178,28 @@ Always run `prettier --write` on changed files.
 
 ### Data Layer (`src/lib/db/`)
 
-All persistence uses SQLite through domain-specific modules:
-`core.ts`, `providers.ts`, `models.ts`, `combos.ts`, `apiKeys.ts`, `settings.ts`,
-`backup.ts`, `proxies.ts`, `prompts.ts`, `webhooks.ts`, `detailedLogs.ts`,
-`domainState.ts`, `registeredKeys.ts`, `quotaSnapshots.ts`, `modelComboMappings.ts`,
-`cliToolState.ts`, `encryption.ts`, `readCache.ts`, `secrets.ts`, `stateReset.ts`,
-`contextHandoffs.ts`.
-Schema migrations live in `db/migrations/` and run via `migrationRunner.ts`.
+All persistence uses SQLite through **83 domain-specific modules** in `src/lib/db/`. Top modules:
+
+- 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`
+
+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 15 base tables. Helpers: `rowToCamel`, `encryptConnectionFields`.
+  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**: 21 files (`001_initial_schema.sql` → `021_combo_call_log_targets.sql`).
-  Each migration is idempotent and runs in a transaction.
+- **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.
@@ -158,19 +214,19 @@ Route → CORS preflight → Body validation (Zod) → Optional auth (extractApi
   → 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 (5 providers)                  |
+| 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.
@@ -212,7 +268,7 @@ Zod schemas, and unit tests aligned when editing.
 ### Provider Categories
 
 - **Free** (4): Qoder AI, Qwen Code, Gemini CLI (deprecated), Kiro AI
-- **OAuth** (8): Claude Code, Antigravity, Codex, GitHub Copilot, Cursor, Kimi Coding, Kilo Code, Cline
+- **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,
@@ -280,12 +336,41 @@ Includes request/response translators with helpers for image handling.
 
 ### Services (`open-sse/services/`)
 
-36+ service modules including: `combo.ts` (routing engine), `usage.ts`, `tokenRefresh.ts`,
+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`, and more.
+`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`)
 
@@ -293,8 +378,8 @@ Includes request/response translators with helpers for image handling.
   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** (13): priority, weighted, fill-first, round-robin, P2C, random, least-used,
-  cost-optimized, strict-random, auto, lkgp, context-optimized, context-relay.
+- **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.
 
@@ -306,7 +391,7 @@ Policy engine modules: `policyEngine.ts`, `comboResolver.ts`, `costRules.ts`,
 
 ### MCP Server (`open-sse/mcp-server/`)
 
-29 tools, 3 transports (stdio / SSE / Streamable HTTP). Scoped auth (10 scopes), Zod schemas.
+**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,
@@ -315,10 +400,23 @@ best_combo_for_task, explain_route, get_session_snapshot, db_health_check, sync_
 
 **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,
@@ -327,7 +425,7 @@ handler: async (args) => {...} }`. Zod validates inputs before the handler fires
   `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** (10): Control which tool categories an API key can access. Enforcement happens
+- **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.
@@ -336,7 +434,7 @@ handler: async (args) => {...} }`. Zod validates inputs before the handler fires
 
 JSON-RPC 2.0, SSE streaming, Task Manager with TTL cleanup.
 Agent Card at `/.well-known/agent.json`.
-Skills: `quotaManagement.ts`, `smartRouting.ts`.
+Skills (6): `smartRouting.ts`, `quotaManagement.ts`, `providerDiscovery.ts`, `costAnalysis.ts`, `healthReport.ts`, `listCapabilities.ts`.
 
 #### A2A Internals
 
@@ -389,6 +487,34 @@ MITM proxy capability with certificate management, DNS handling, and target rout
 
 Request middleware including `promptInjectionGuard.ts`.
 
+### Guardrails (`src/lib/guardrails/`)
+
+Hot-reloadable guardrails framework (3 built-in: pii-masker, prompt-injection, vision-bridge). Fail-open. The `pii-masker` guardrail is registered and runs on every request, but its data-mutating logic is **opt-in** and OFF by default — it only redacts when `PII_REDACTION_ENABLED` (request) / `PII_RESPONSE_SANITIZATION` (response + streaming) are enabled (both `defaultValue: "false"`); with them off, payloads pass through untouched. A request can additionally opt OUT of any guardrail via header (`x-omniroute-disabled-guardrails`). Never make PII default-on (Hard Rule #20). 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`
@@ -401,10 +527,59 @@ Request middleware including `promptInjectionGuard.ts`.
 
 ## Subdirectory AGENTS.md Files
 
-- **[`open-sse/AGENTS.md`](open-sse/AGENTS.md)** — Streaming engine, request pipeline, handlers, and executors
 - **[`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
@@ -418,3 +593,4 @@ Request middleware including `promptInjectionGuard.ts`.
 - **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

@@ -1,7 +1,6 @@
-# CLAUDE.md — AI Agent Session Bootstrap
+# CLAUDE.md
 
-> Quick-start context for AI coding agents. For deep architecture details, see `AGENTS.md`.
-> For contribution workflow, see `CONTRIBUTING.md`.
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
 
 ## Quick Start
 
@@ -12,60 +11,52 @@ 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% min)
+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 a Single Test
+### Running Tests
 
 ```bash
-# Node.js native test runner (most tests)
-node --import tsx/esm --test tests/unit/your-file.test.mjs
+# 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, 100+ LLM providers, auto-fallback.
+**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)   |
-| Services        | `open-sse/services/`     | Combo routing, rate limits, caching, etc   |
-| Database        | `src/lib/db/`            | SQLite domain modules (22 files)           |
-| Domain/Policy   | `src/domain/`            | Policy engine, cost rules, fallback logic  |
-| MCP Server      | `open-sse/mcp-server/`   | 25 tools, 3 transports, 10 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           |
-| UI Components   | `src/shared/components/` | React components (Tailwind CSS v4)         |
-| Provider Consts | `src/shared/constants/`  | Provider registry (Zod-validated)          |
-| Validation      | `src/shared/validation/` | Zod v4 schemas                             |
-| Tests           | `tests/`                 | Unit, integration, e2e, security, load     |
+| 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 Layout
-
-```
-OmniRoute/              # Root package
-├── src/                # Next.js 16 app (TypeScript)
-├── open-sse/           # @omniroute/open-sse workspace (streaming engine)
-├── electron/           # Desktop app (Electron)
-├── tests/              # All test suites
-├── docs/               # Documentation
-└── bin/                # CLI entry point
-```
+Monorepo: `src/` (Next.js 16 app), `open-sse/` (streaming engine workspace), `electron/` (desktop app), `tests/`, `bin/` (CLI entry point).
 
 ---
 
-## Request Pipeline (Abbreviated)
+## Request Pipeline
 
 ```
 Client → /v1/chat/completions (Next.js route)
@@ -76,39 +67,187 @@ Client → /v1/chat/completions (Next.js route)
     → 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`): 17 strategies (priority, weighted, fill-first, round-robin, P2C, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion). Each target calls `handleSingleModel()` which wraps `handleChatCore()` with per-target error handling and circuit breaker checks. The `fusion` strategy is the exception: it fans out to a panel of models in parallel, then a judge model synthesizes one final answer (`open-sse/services/fusion.ts`). See `docs/routing/AUTO-COMBO.md` for the 9-factor Auto-Combo scoring + the full strategy table 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
+- **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 Access
+### Database
 
-- **Always** go through `src/lib/db/` domain modules
-- **Never** write raw SQL in routes or handlers
+- **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
+- **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/` — 21 versioned SQL files
+- 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
+- Never swallow errors in SSE streams — use abort signals for cleanup
 - Return proper HTTP status codes (4xx/5xx)
 
 ### Security
 
-- **Never** commit secrets/credentials
 - **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.
 
 ---
 
@@ -117,11 +256,11 @@ Client → /v1/chat/completions (Next.js route)
 ### 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
+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
+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/` (registration, translation, error handling)
+6. Write tests in `tests/unit/` (include the publicCreds shape assertion if you added a new embedded default)
 
 ### Adding a New API Route
 
@@ -129,53 +268,124 @@ Client → /v1/chat/completions (Next.js 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. Add tests
+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`
-2. Import `getDbInstance` from `./core.ts`
-3. Export CRUD functions for your domain table(s)
-4. Add migration in `src/lib/db/migrations/` if new tables needed
-5. Re-export from `src/lib/localDb.ts` (add to the re-export list only)
-6. Write tests
+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/`
-2. Define Zod input schema + async handler
-3. Register in tool set (wired by `createMcpServer()`)
-4. Assign to appropriate scope(s)
-5. Write tests (tool invocation logged to `mcp_audit` table)
+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/`
+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 the DB-backed skill registry
-4. Write tests
+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`
 
 ---
 
-## Testing Cheat Sheet
+## Reference Documentation
 
-| What                    | Command                                                 |
-| ----------------------- | ------------------------------------------------------- |
-| All tests               | `npm run test:all`                                      |
-| Unit tests              | `npm run test:unit`                                     |
-| Single file             | `node --import tsx/esm --test tests/unit/file.test.mjs` |
-| 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% min all metrics)           |
-| Coverage report         | `npm run coverage:report`                               |
+For any non-trivial change, read the matching deep-dive first:
 
-**PR rule**: If you change production code in `src/`, `open-sse/`, `electron/`, or `bin/`,
-you must include or update tests in the same PR.
+| 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, 17 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
@@ -183,40 +393,106 @@ you must include or update tests in the same PR.
 ```bash
 # Never commit directly to main
 git checkout -b feat/your-feature
-# ... make changes ...
 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](https://www.conventionalcommits.org/)):
+**Commit format** (Conventional Commits): `feat(db): add circuit breaker` — scopes: `db`, `sse`, `oauth`, `dashboard`, `api`, `cli`, `docker`, `ci`, `mcp`, `a2a`, `memory`, `skills`
 
-```
-feat: add circuit breaker for provider calls
-fix: resolve JWT secret validation edge case
-docs: update AGENTS.md with pipeline internals
-test: add MCP tool unit tests
-refactor(db): consolidate rate limit tables
-```
+**Husky hooks**:
 
-**Scopes**: `db`, `sse`, `oauth`, `dashboard`, `api`, `cli`, `docker`, `ci`, `mcp`, `a2a`,
-`memory`, `skills`.
+- **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).
+   **🔴 MANDATORY PATH: every worktree lives under `.claude/worktrees/` — and nowhere else.**
+   This is the single canonical location (the same dir the native `EnterWorktree` tool uses). It
+   is gitignored AND in the `tsconfig.json` / `.dockerignore` excludes, so worktrees never leak
+   into the build scope. **Never** use `.worktrees/`, repo-root, or any other path — a worktree
+   outside `.claude/worktrees/` (a) escapes the build-scope excludes and poisons `next build` (the
+   `tsconfig` `include: **/*` globs ~70× the codebase → OOM; incident 2026-06-25) and (b) scatters
+   worktrees across two dirs.
+
+   ```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 "$BASE_BRANCH"
+   git worktree add ".claude/worktrees/${TASK##*/}" -b "$TASK" "origin/$BASE_BRANCH"
+   cd ".claude/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 (it already creates worktrees under
+   `.claude/worktrees/`): 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 .claude/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 ≥18 <24, ES Modules
-- **TypeScript**: 5.9, target ES2022, module esnext, resolution bundler
-- **Path aliases**: `@/*` → `src/`, `@omniroute/open-sse` → `open-sse/`
+- **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`, `INITIAL_PASSWORD`, `REQUIRE_API_KEY`, `APP_LOG_LEVEL`
+- **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`)
 
 ---
 
-## Hard Rules (Never Violate)
+## 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`
@@ -226,4 +502,31 @@ refactor(db): consolidate rate limit tables
 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 stay ≥60% (statements, lines, functions, branches)
+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".
+20. PII redaction/sanitization is **opt-in — never on by default**. OmniRoute proxies for self-hosted/local LLMs where the operator owns the data, so mutating request/response payloads by default would silently corrupt legitimate traffic. The two data-mutating PII feature flags **MUST** keep `defaultValue: "false"` in `src/shared/constants/featureFlagDefinitions.ts`: `PII_REDACTION_ENABLED` (request-side) and `PII_RESPONSE_SANITIZATION` (response + streaming). All three application points — `src/lib/guardrails/piiMasker.ts` (request guardrail), `src/lib/piiSanitizer.ts` (response), `src/lib/streamingPiiTransform.ts` (SSE) — are gated on these flags; with both off the `pii-masker` guardrail still runs but never mutates payloads (data passes through untouched). Flipping either default to `"true"` requires explicit operator approval. The regression guard is `tests/unit/pii-opt-in-default.test.ts` (asserts both definition defaults + behavioral pass-through). Opt-in is per-operator via env or the settings/DB override (`src/lib/db/featureFlags.ts`), never a silent default. See `docs/security/GUARDRAILS.md`.
+
+---
+
+## 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

@@ -17,23 +17,23 @@ diverse, inclusive, and healthy community.
 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,
+- 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
+- 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
+- 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
+- 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
+- Other conduct which could reasonably be considered inappropriate in a
   professional setting
 
 ## Enforcement Responsibilities
@@ -59,8 +59,11 @@ 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 at
-.
+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
@@ -106,7 +109,7 @@ 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
+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
@@ -115,8 +118,8 @@ the community.
 ## Attribution
 
 This Code of Conduct is adapted from the [Contributor Covenant][homepage],
-version 2.0, available at
-https://www.contributor-covenant.org/version/2/0/code_of_conduct.html.
+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).

CONTRIBUTING.md

@@ -8,7 +8,7 @@ Thank you for your interest in contributing! This guide covers everything you ne
 
 ### Prerequisites
 
-- **Node.js** >= 18 < 24 (recommended: 22 LTS)
+- **Node.js** `>=22.22.3 <23`, or `>=24.0.0 <27` (recommended: 24 LTS)
 - **npm** 10+
 - **Git**
 
@@ -59,13 +59,40 @@ These settings are stored in the database and persist across restarts, overridin
 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:20128/dashboard`
@@ -108,7 +135,7 @@ test: add observability unit tests
 refactor(db): consolidate rate limit tables
 ```
 
-Scopes: `db`, `sse`, `oauth`, `dashboard`, `api`, `cli`, `docker`, `ci`, `mcp`, `a2a`, `memory`, `skills`.
+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`.
 
 ---
 
@@ -133,23 +160,35 @@ npm run test:protocols:e2e
 # Ecosystem compatibility tests
 npm run test:ecosystem
 
-# Coverage (60% min statements/lines/functions/branches)
+# 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
+
+# Gated real-upstream combo smoke (requires VPS access + real provider credits)
+# Hits REAL providers — costs a little. NEVER runs in CI. Skips cleanly without the gate.
+# Needs: ssh root@192.168.0.15 access (sources a read-only DB snapshot from the VPS).
+RUN_COMBO_LIVE=1 npm run test:combo:live
+
+# Phase-3 VPS live smoke — plain Node ESM scripts, hit the live .15 server directly.
+# Requires: ssh root@192.168.0.15 access (combos created/torn down via SSH sqlite).
+# Hits REAL providers (small cost). Creates/deletes only __live_test__* combos. NEVER runs in CI.
+# REQUIRE_API_KEY=false on .15 so no API key needed, but honors COMBO_LIVE_BASE_URL / COMBO_LIVE_API_KEY if set.
+npm run test:combo:live:vps              # 7 HTTP scenarios (priority/round-robin/weighted/cost/fusion/auto + health)
+npm run test:combo:live:vps:failover     # adds a real cross-provider failover scenario (8 total)
 ```
 
 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 overall coverage gate at **60% or higher** for statements, lines, functions, and branches
+- 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/COVERAGE_PLAN.md` for the phased coverage improvement roadmap
+- See `docs/ops/COVERAGE_PLAN.md` for the phased coverage improvement roadmap
 
 ### Pull Request Requirements
 
@@ -157,7 +196,7 @@ Before opening or merging a PR:
 
 - Run `npm run test:unit`
 - Run `npm run test:coverage`
-- Ensure the coverage gate stays at **60%+** for all metrics
+- 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
 
@@ -208,7 +247,7 @@ src/                        # TypeScript (.ts / .tsx)
 ├── mitm/                   # MITM proxy (cert, DNS, target routing)
 ├── shared/
 │   ├── components/         # React components (.tsx)
-│   ├── constants/          # Provider definitions (60+), MCP scopes, routing strategies
+│   ├── constants/          # Provider definitions (177), MCP scopes, 14 routing strategies
 │   ├── utils/              # Circuit breaker, sanitizer, auth helpers
 │   └── validation/         # Zod v4 schemas
 └── sse/                    # SSE proxy pipeline
@@ -225,25 +264,32 @@ open-sse/                   # @omniroute/open-sse workspace
 electron/                   # Electron desktop app (cross-platform)
 
 tests/
-├── unit/                   # Node.js test runner (122 test files)
+├── 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
-├── ARCHITECTURE.md         # System architecture
-├── API_REFERENCE.md        # All endpoints
-├── USER_GUIDE.md           # Provider setup, CLI integration
-├── TROUBLESHOOTING.md      # Common issues
-├── MCP-SERVER.md           # MCP server (25 tools)
-├── A2A-SERVER.md           # A2A agent protocol
-├── AUTO-COMBO.md           # Auto-combo engine
-├── CLI-TOOLS.md            # CLI tools integration
-├── COVERAGE_PLAN.md        # Test coverage improvement plan
-├── openapi.yaml            # OpenAPI specification
-└── 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
 ```
 
 ---
@@ -266,6 +312,10 @@ Create request/response translators in `open-sse/translator/`.
 
 Add OAuth credentials in `src/lib/oauth/constants/oauth.ts` and service in `src/lib/oauth/services/`.
 
+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).
+
+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 5: Register Models
 
 Add model definitions in `open-sse/config/providerRegistry.ts`.
@@ -287,9 +337,15 @@ 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)
 
 ---
 
@@ -297,11 +353,17 @@ Write unit tests in `tests/unit/` covering at minimum:
 
 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.
 
+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)
-- **API Reference**: See [`docs/API_REFERENCE.md`](docs/API_REFERENCE.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

@@ -1,22 +1,74 @@
-FROM node:24.15.0-trixie-slim AS builder
+# ── Common base with runtime deps ──────────────────────────────────────────
+FROM node:24-trixie-slim AS base
 WORKDIR /app
 
-RUN apt-get update \
+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 ./
-COPY scripts/postinstall.mjs ./scripts/postinstall.mjs
-COPY scripts/postinstallSupport.mjs ./scripts/postinstallSupport.mjs
-COPY scripts/native-binary-compat.mjs ./scripts/native-binary-compat.mjs
+# Workspace package manifests MUST be present before `npm ci` so npm materializes
+# the workspace and installs its *workspace-only* deps (e.g. safe-regex,
+# @toon-format/toon — declared in open-sse/package.json, not hoisted to root).
+# Without this, `npm ci` skips them and `npm run build` fails with "Module not
+# found" (root cause of the v3.8.39 Docker build break). workspaces = ["open-sse"].
+COPY open-sse/package.json ./open-sse/package.json
+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
-RUN if [ -f package-lock.json ]; then npm ci --no-audit --no-fund; else npm install --no-audit --no-fund; fi
+# --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 -- --webpack
+RUN --mount=type=cache,target=/app/.build/next/cache \
+  mkdir -p /app/data && npm run build
 
-FROM node:24.15.0-trixie-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" \
@@ -27,49 +79,113 @@ LABEL org.opencontainers.image.title="omniroute" \
 ENV NODE_ENV=production
 ENV PORT=20128
 ENV HOSTNAME=0.0.0.0
-ENV NODE_OPTIONS="--max-old-space-size=256"
+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 apt-get update \
-  && apt-get install -y --no-install-recommends libsecret-1-0 ca-certificates \
-  && rm -rf /var/lib/apt/lists/*
 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 ./
-# Explicitly copy @swc/helpers — not always traced by standalone output but needed at runtime
-COPY --from=builder /app/node_modules/@swc/helpers ./node_modules/@swc/helpers
-# Explicitly copy pino transport dependencies — pino spawns a worker that requires
-# pino-abstract-transport at runtime; Next.js standalone trace does not capture it (#449)
-COPY --from=builder /app/node_modules/pino-abstract-transport ./node_modules/pino-abstract-transport
-COPY --from=builder /app/node_modules/pino-pretty ./node_modules/pino-pretty
-COPY --from=builder /app/node_modules/split2 ./node_modules/split2
-# Migration SQL files are read via fs.readFileSync at runtime and are NOT
-# traced by Next.js standalone output — copy them explicitly.
-COPY --from=builder /app/src/lib/db/migrations ./migrations
+# `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
 
-COPY --from=builder /app/scripts/run-standalone.mjs ./run-standalone.mjs
-COPY --from=builder /app/scripts/runtime-env.mjs ./runtime-env.mjs
-COPY --from=builder /app/scripts/bootstrap-env.mjs ./bootstrap-env.mjs
-COPY --from=builder /app/scripts/healthcheck.mjs ./healthcheck.mjs
+# 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
 
+# 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
+
+# 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", "run-standalone.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 \
+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

@@ -1,21 +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 the `scripts/` directory or `scripts/scratch/` for temporary one-offs. NEVER dump loose scripts 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`, etc.)
+- 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`, `AGENTS.md`)
-- CI/CD files and ignore definitions (`.gitignore`, `.dockerignore`)
+- 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/scratch/` or the `tests/unit/` directories according to your goals. Do not pollute the `/` root context.
+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. VPS Dashboard Credentials
+## 2. Hard Rules (mirror of `CLAUDE.md`)
 
-| Environment | URL                       | Password |
-| ----------- | ------------------------- | -------- |
-| Local VPS   | http://192.168.0.15:20128 | 123456   |
+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.