Merge remote-tracking branch 'origin/release/v3.8.50' into feat/9570-plugin-context-headers

Closes #9030

.claude/worktrees/feat-7786/changelog.d/features/7786-management-auth-terminology-docs.md

@@ -0,0 +1 @@
+- **docs:** add management authentication terminology guide ([#7786](https://github.com/diegosouzapw/OmniRoute/issues/7786))

.claude/worktrees/feat-7786/docs/guides/MANAGEMENT-AUTH.md

@@ -0,0 +1,41 @@
+# Management Authentication
+
+OmniRoute uses four distinct credential families for management access. This guide
+distinguishes them by purpose, scope, and locality.
+
+| Credential              | Scope              | Locality      | Use Case                          |
+|-------------------------|--------------------|---------------|-----------------------------------|
+| Dashboard JWT session   | Full management    | Localhost      | Web dashboard login               |
+| CLI machine-id token    | Full management    | Per-machine    | `omniroute` CLI commands          |
+| Scoped `oma_` token     | Configurable scope | External       | Automation / CI / API access      |
+| Manage-scope API key    | `manage` scope     | External       | Management API calls              |
+
+## Dashboard JWT Session
+
+Generated on dashboard login (`/api/auth/login`). Stored in HTTP-only cookie.
+Valid for the session duration. Cannot be used from external hosts.
+
+## CLI Machine-ID Token
+
+Created by `omniroute auth login` on first use. Stored in `~/.omniroute/auth.json`.
+Used by the CLI for all management operations. Tied to the machine identity.
+
+## Scoped `oma_` Access Token
+
+Created via dashboard or CLI with configurable scopes (e.g., `manage`, `read`).
+Format: `oma_<random-hex>`. Used for programmatic access from external systems.
+
+## Manage-Scope API Key
+
+Standard API key with the `manage` scope enabled. Created in dashboard API Keys page.
+Used for management API calls from external hosts.
+
+## Header Examples
+
+```
+Authorization: Bearer oma_abc123def456
+Authorization: Bearer <standard-api-key-with-manage-scope>
+Cookie: omniroute_session=<jwt-token>
+```
+
+See `docs/reference/API_REFERENCE.md` for endpoint-specific auth requirements.

.claude/worktrees/feat-7786/tests/unit/management-auth-docs.test.ts

@@ -0,0 +1,27 @@
+import { describe, it } from "node:test";
+import { ok } from "node:assert/strict";
+import { readFileSync } from "node:fs";
+
+describe("Management auth documentation (#7786)", () => {
+  const docPath = "docs/guides/MANAGEMENT-AUTH.md";
+  const content = readFileSync(docPath, "utf-8");
+
+  it("exists and has content", () => {
+    ok(content.length > 500, "should have substantial content");
+    ok(content.includes("Dashboard JWT session"));
+    ok(content.includes("CLI machine-id token"));
+    ok(content.includes("oma_"));
+  });
+
+  it("documents all four credential families", () => {
+    const families = ["Dashboard JWT", "CLI machine-id", "oma_", "Manage-scope"];
+    for (const f of families) {
+      ok(content.includes(f), `should document ${f}`);
+    }
+  });
+
+  it("mentions relevant auth header examples", () => {
+    ok(content.includes("Authorization"));
+    ok(content.includes("Bearer"));
+  });
+});

.claude/worktrees/feat-8635/changelog.d/features/8635-systemd-autostart-linux.md

@@ -0,0 +1 @@
+- **feat(infra):** add systemd autostart unit for Linux ([#8635](https://github.com/diegosouzapw/OmniRoute/issues/8635))

.claude/worktrees/feat-8635/contrib/systemd/omniroute.service

@@ -0,0 +1,19 @@
+[Unit]
+Description=OmniRoute AI Proxy
+After=network.target network-online.target
+Wants=network-online.target
+
+[Service]
+Type=simple
+ExecStart=$(which omniroute) start
+Restart=on-failure
+RestartSec=5
+Environment=NODE_ENV=production
+
+# Security hardening
+NoNewPrivileges=true
+ProtectSystem=full
+PrivateTmp=true
+
+[Install]
+WantedBy=default.target

.claude/worktrees/feat-8635/tests/unit/systemd-autostart.test.ts

@@ -0,0 +1,23 @@
+import { describe, it } from "node:test";
+import { ok } from "node:assert/strict";
+import { readFileSync, existsSync } from "node:fs";
+
+describe("Systemd autostart (#8635)", () => {
+  const svcPath = "contrib/systemd/omniroute.service";
+  const content = readFileSync(svcPath, "utf-8");
+
+  it("service file exists", () => {
+    ok(existsSync(svcPath));
+    ok(content.length > 200);
+  });
+
+  it("defines required systemd sections", () => {
+    ok(content.includes("[Unit]"));
+    ok(content.includes("[Service]"));
+    ok(content.includes("[Install]"));
+  });
+
+  it("specifies WantedBy=default.target", () => {
+    ok(content.includes("WantedBy=default.target"));
+  });
+});

.dockerignore

@@ -24,6 +24,7 @@ coverage
 # Runtime data and logs
 data
 logs
+.sandbox
 
 # Local env files (inject at runtime via --env-file or -e)
 .env

.env.devin-bridge.example

@@ -0,0 +1,6 @@
+ENABLE_LIVE_DEVIN_TESTS=0
+DEVIN_BRIDGE_MODEL=devin-cli-agentic/swe-1-7
+DEVIN_BRIDGE_SONNET_MODEL=devin-cli-agentic/swe-1-7
+DEVIN_BRIDGE_OPUS_MODEL=devin-cli-agentic/swe-1-7
+DEVIN_BRIDGE_HAIKU_MODEL=devin-cli-agentic/swe-1-7
+DEVIN_BRIDGE_SUBAGENT_MODEL=devin-cli-agentic/swe-1-7

.env.example

@@ -353,6 +353,7 @@ ALLOW_API_KEY_REVEAL=false
 # instead of growing an unbounded string until the V8 heap is exhausted.
 # Used by: open-sse/handlers/chatCore/nonStreamingResponseBody.ts
 # Default: 67108864 (64 MB)
+# OMNIROUTE_FORWARDING_HEADER_BUDGET_BYTES=768
 # OMNIROUTE_MAX_NONSTREAMING_RESPONSE_BYTES=67108864
 
 # CORS configuration — controls which cross-origin browser clients can call the API.
@@ -857,6 +858,12 @@ PROVIDER_LIMITS_SYNC_SPACING_MS=1500
 # (>= 3 retrievals = never compressed). 1 disables the ramp (binary skip at the threshold only).
 # Used by: open-sse/services/compression/engines/ccr/index.ts.  Default: 2.
 #COMPRESSION_CCR_RETRIEVAL_RAMP_FACTOR=2
+# CCR durable block store (#9061). The in-memory store loses blocks to LRU eviction, the TTL, a
+# restart, or a retrieve landing on another instance, while the model is told it can retrieve them
+# verbatim. Set to false to keep blocks in memory only, at the cost of that promise. Blocks over
+# 512KB and cloud runtimes are memory-only regardless.
+# Used by: open-sse/services/compression/engines/ccr/index.ts.  Default: true.
+#COMPRESSION_CCR_DURABLE_STORE=true
 # T08/H5 — usage-observed prefix freeze (OPT-IN, default off). When enabled, a system prompt seen
 # >= THRESHOLD times is treated as a stable cacheable prefix and preserved from compression even
 # for providers the static cache-aware heuristic does not recognize (freeze = preserve, never
@@ -1041,6 +1048,17 @@ GITHUB_OAUTH_CLIENT_ID=Iv1.b507a08c87ecfe98
 # VISION_BRIDGE_BASE_URL=
 # VISION_BRIDGE_API_KEY=
 
+# ── Raycast Pro (local auto-import) ──
+# Raycast Pro AI is a reverse-engineered, unofficial API — local/personal use
+# only (no OAuth client_id/secret; token is captured via macOS Auto-Import
+# from the Keychain + local Raycast SQLite DB, or pasted manually). These
+# vars are optional manual overrides used by open-sse/services/raycast.ts
+# and the direct-probe benchmark script scripts/raycast/usage-benchmark.mjs.
+# RAYCAST_BEARER_TOKEN=
+# RAYCAST_DEVICE_ID=
+# RAYCAST_AID=
+# RAYCAST_SIG_SECRET=
+
 # ─────────────────────────────────────────────────────────────────────────────
 # ⚠️  GOOGLE OAUTH (Antigravity) & OTHER PROVIDERS — REMOTE SERVERS
 # ─────────────────────────────────────────────────────────────────────────────
@@ -1253,6 +1271,14 @@ CURSOR_USER_AGENT="Cursor/3.4"
 # OMNIROUTE_BROWSER_POOL=on
 # WEB_COOKIE_USE_BROWSER=0
 
+# ── Adobe Firefly browser sign-in (system Chrome/Edge CDP) ──
+# Used by: open-sse/services/adobeFireflyBrowserLogin.ts. The Firefly login
+# flow drives a real, system-installed Chrome or Microsoft Edge via CDP so the
+# user can sign in interactively; the executable is auto-detected from common
+# install paths per OS. Set this to override that detection (e.g. a portable
+# install or a non-standard path) when auto-detection fails.
+# OMNIROUTE_LOGIN_BROWSER_PATH=
+
 # ── Circuit breaker thresholds and reset windows ──
 # Used by: open-sse/config/constants.ts → src/lib/resilience/settings.ts.
 # Defaults match historical PROVIDER_PROFILES values (post-scaling for
@@ -1523,6 +1549,14 @@ APP_LOG_TO_FILE=true
 # Default: 86400000 (24 hours)
 # OPENROUTER_CATALOG_TTL_MS=86400000
 
+# Enrich the dashboard providers list with OpenRouter weekly ranking stats.
+# ON by default; set false to skip the background fetch entirely (#9324).
+# Used by: src/lib/catalog/openrouterProviderStats.ts
+# OPENROUTER_PROVIDER_STATS_ENABLED=true
+# Cache TTL for the OpenRouter provider stats snapshot, in ms.
+# Default: 86400000 (24 hours)
+# OPENROUTER_PROVIDER_STATS_TTL_MS=86400000
+
 # ── Model catalog response shape ──
 # Include display-friendly name fields in /v1/models responses.
 # Disable for clients that expect model IDs only.
@@ -1543,6 +1577,13 @@ APP_LOG_TO_FILE=true
 # DESIGNER_WEB_POLL_TIMEOUT_MS=60000   # Max wait for job completion (default: 60s)
 # DESIGNER_WEB_POLL_INTERVAL_MS=2000   # Poll frequency (default: 2s)
 
+# ── Adobe Firefly (Image Upscale) ──
+# Base delay (ms) for the submit-retry exponential backoff when Adobe Firefly's
+# upscale job submission is rate-limited. Used by:
+# open-sse/services/adobeFireflyUpscale.ts::submitRetryDelayMs.
+# Default: 8000 (20 under NODE_ENV=test/VITEST/NODE_TEST_CONTEXT).
+# ADOBE_FIREFLY_SUBMIT_BASE_DELAY_MS=8000
+
 # ── AWS Bedrock (Kiro / Audio) ──
 # Region used to construct AWS Bedrock endpoints. Used by:
 # src/lib/providers/validation.ts and open-sse/handlers/audioSpeech.ts.
@@ -1637,6 +1678,26 @@ APP_LOG_TO_FILE=true
 # Used by: src/lib/services/bootstrap.ts, src/app/api/services/mux/_lib.ts
 # MUX_SERVICE_PORT=8322
 
+# ── Dario embedded service ──
+# Override the host/port the embedded Dario (Claude Code subscription proxy)
+# daemon binds to and is reached at. Always bound to 127.0.0.1 — never
+# configurable to 0.0.0.0. Rarely needed — defaults to 127.0.0.1:3456.
+# Used by: src/lib/services/installers/dario.ts, src/lib/services/bootstrap.ts,
+# src/app/api/services/dario/_lib.ts, src/app/api/services/dario/admin/_lib.ts,
+# open-sse/executors/dario.ts
+# DARIO_HOST=127.0.0.1
+# DARIO_PORT=3456
+
+# ── Dario embedded service ──
+# Override the host/port the embedded Dario (Claude Code subscription proxy)
+# daemon binds to and is reached at. Always bound to 127.0.0.1 — never
+# configurable to 0.0.0.0. Rarely needed — defaults to 127.0.0.1:3456.
+# Used by: src/lib/services/installers/dario.ts, src/lib/services/bootstrap.ts,
+# src/app/api/services/dario/_lib.ts, src/app/api/services/dario/admin/_lib.ts,
+# open-sse/executors/dario.ts
+# DARIO_HOST=127.0.0.1
+# DARIO_PORT=3456
+
 # ── Local hostnames (Docker networking) ──
 # Comma-separated additional hostnames treated as "local" for provider routing.
 # Used by: open-sse/config/providerRegistry.ts — allows Docker service names.
@@ -1869,6 +1930,18 @@ APP_LOG_TO_FILE=true
 # ── Devin CLI binary path ──
 # Used by: open-sse/executors/devin-cli.ts. Default: looked up via PATH.
 # CLI_DEVIN_BIN=devin
+# Agentic bridge-only binary override. The bridge still executes ACP stdio only.
+# CLI_DEVIN_AGENTIC_BIN=devin
+# Required isolated HOME for the agentic Devin child process.
+# DEVIN_AGENTIC_HOME=/home/bridge
+# Bounded ACP turn timeout in milliseconds. Default: 120000.
+# DEVIN_AGENTIC_ACP_TIMEOUT_MS=120000
+# Agentic bridge model aliases. Values must keep the devin-cli-agentic/ prefix.
+# DEVIN_BRIDGE_MODEL=devin-cli-agentic/swe-1-7
+# DEVIN_BRIDGE_SONNET_MODEL=devin-cli-agentic/swe-1-7
+# DEVIN_BRIDGE_OPUS_MODEL=devin-cli-agentic/swe-1-7
+# DEVIN_BRIDGE_HAIKU_MODEL=devin-cli-agentic/swe-1-7
+# DEVIN_BRIDGE_SUBAGENT_MODEL=devin-cli-agentic/swe-1-7
 
 # ── Command Code (custom CLI) callback ──
 # Local port used for OAuth-style callbacks from the Command Code CLI helper.
@@ -2139,6 +2212,11 @@ PLAYGROUND_COMPARE_MAX_COLUMNS=4
 # MEMORY_TYPED_DECAY_EPISODIC_DAYS=30           # episodic TTL in days; 0 = episodic immune too
 # MEMORY_TYPED_DECAY_ACCESS_IMMUNITY=3          # access_count >= N → immune; 0 disables access immunity
 # MEMORY_TYPED_DECAY_SWEEP_INTERVAL=0           # periodic sweep interval (seconds); 0 = no periodic sweep
+# ─── Memory Backend Connectors (Generic HTTP) ──────────────────────────────
+# NOTION_API_KEY=
+# NOTION_API_URL=
+# OBSIDIAN_API_KEY=
+# OBSIDIAN_API_URL=
 # AgentBridge + Traffic Inspector (Group A)
 
 # AgentBridge
@@ -2154,6 +2232,15 @@ INSPECTOR_MAX_BODY_KB=1024
 INSPECTOR_MASK_SECRETS=true
 INSPECTOR_LLM_HOSTS_EXTRA=
 INSPECTOR_INTERNAL_INGEST_TOKEN=
+# Shared secret for identity-preserving internal REST hops (#9260): when an
+# OmniRoute component calls another local OmniRoute route, this token (sent as
+# x-omniroute-internal-service-token) marks the request as internal so the
+# original caller identity is preserved. OPT-IN: unset disables the mechanism.
+# Used by: src/lib/api/internalServiceAuth.ts
+# OMNIROUTE_INTERNAL_SERVICE_TOKEN=
+# File-based variant (secret-file pattern; wins only when the inline var is
+# unset): path to a file whose trimmed content is the token.
+# OMNIROUTE_INTERNAL_SERVICE_TOKEN_FILE=
 # Quota Sharing (Group B — planos 16+22)
 QUOTA_STORE_DRIVER=sqlite              # sqlite | redis
 # QUOTA_STORE_REDIS_URL=               # ex.: redis://localhost:6379 (apenas quando driver=redis)
@@ -2370,3 +2457,29 @@ QUOTA_STORE_DRIVER=sqlite              # sqlite | redis
 # OMNIROUTE_DATA_DIR are both unset. Locates the Notion web-thread session cache.
 # ─────────────────────────────────────────────────────────────────────────────
 # VIBEPROXY_DATA_DIR=
+
+# ── Internal service auth (management-plane service-to-service calls) ─────────
+# Inline token for internal service authentication; prefer the _FILE variant in
+# containerized deployments so the secret never lands in the environment table.
+# OMNIROUTE_INTERNAL_SERVICE_TOKEN=
+# Path to a file containing the internal service token (overrides the inline var).
+# OMNIROUTE_INTERNAL_SERVICE_TOKEN_FILE=
+
+# ═══════════════════════════════════════════════════════════════════════════════
+# 26. RADAR FEED (SELF-HOSTING)
+# ═══════════════════════════════════════════════════════════════════════════════
+# Optional add-on (feature flag RADAR_ENABLED, default off — see feature flag
+# settings, not an env var) that overlays a signed, freshly-curated free-model
+# catalog on top of the release baseline. Both variables below are optional and
+# only needed to point the client at a self-hosted/forked feed instead of the
+# default OmniRoute Radar feed. Used by: src/lib/radar/sync.ts,
+# src/lib/radar/pinnedKeys.ts.
+
+# Base URL of the Radar feed service. Overrides the built-in default so forks
+# and self-hosters can point at their own signed feed.
+# RADAR_FEED_URL=https://radar.omniroute.online
+
+# Ed25519 public key (base64-DER SPKI or PEM) used to verify the feed
+# signature, replacing the pinned default key. Required when self-hosting a
+# feed signed with a different key pair.
+# RADAR_FEED_PUBKEY=

.github/workflows/ci.yml

@@ -1213,8 +1213,10 @@ jobs:
           cache: npm
       - uses: ./.github/actions/npm-ci-retry
       - run: npm run check:node-runtime
-      # (tsx/esm = QW-b; o alinhamento de ESCOPO do integration com o npm script fica p/ follow-up)
-      - run: node --import tsx/esm --import ./tests/_setup/isolateDataDir.ts --test --test-force-exit --test-concurrency=1 --test-shard=${{ matrix.shard }}/2 tests/integration/*.test.ts
+      - name: Integration tests (shard ${{ matrix.shard }}/2)
+        env:
+          TEST_SHARD: ${{ matrix.shard }}/2
+        run: npm run test:integration:ci
 
   test-security:
     name: Security Tests

.github/workflows/docker-publish.yml

@@ -4,6 +4,7 @@ on:
   push:
     branches:
       - main
+      - "release/v*"
     tags:
       - "v*"
     paths-ignore:
@@ -57,39 +58,20 @@ jobs:
           REF_TYPE: ${{ github.ref_type }}
           INPUT_VERSION: ${{ inputs.version }}
           PROMOTE_INPUT: ${{ inputs.promote_latest }}
+          DEFAULT_BRANCH: ${{ github.event.repository.default_branch }}
         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: $VERSION" >&2
-            exit 1
-          fi
+          # 1) Resolve version/channel from the trigger. Only the current default
+          # release branch publishes the mutable `next` channel; main keeps `main`.
+          VERSION=$(bash scripts/ci/resolve-docker-publish-version.sh \
+            "$EVENT_NAME" "$REF_TYPE" "$REF_NAME" "$INPUT_VERSION" "$DEFAULT_BRANCH")
           echo "version=$VERSION" >> "$GITHUB_OUTPUT"
 
-          # 2) Decide whether to promote :latest.
+          # 2) Decide whether to promote :latest. Floating channels are never
+          # eligible, and the helper independently fails closed for non-semver.
           PROMOTE="false"
-          if [ "$VERSION" = "main" ]; then
+          if [ "$VERSION" = "main" ] || [ "$VERSION" = "next" ]; then
             PROMOTE="false"
           elif printf '%s' "$VERSION" | grep -qE -- '-(rc|alpha|beta|pre|next)'; then
             echo "Pre-release identifier detected — skipping :latest."
@@ -109,10 +91,10 @@ jobs:
           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).
+          # 3) Skip immutable version tags that already exist. Floating `main`
+          # and `next` channels are intentionally rebuilt on every matching push.
           SKIP="false"
-          if [ "$VERSION" != "main" ]; then
+          if [ "$VERSION" != "main" ] && [ "$VERSION" != "next" ]; then
             if docker manifest inspect "diegosouzapw/omniroute:${VERSION}" >/dev/null 2>&1; then
               echo "Image diegosouzapw/omniroute:${VERSION} already exists on Docker Hub — skipping rebuild."
               SKIP="true"
@@ -397,7 +379,7 @@ jobs:
 
       - name: Update Docker Hub description
         # Only refresh README/description when we actually promote :latest
-        # (avoids overwriting from main pushes or back-fill builds).
+        # (avoids overwriting from main, next, or back-fill builds).
         if: needs.prepare.outputs.promote_latest == 'true'
         uses: peter-evans/dockerhub-description@v5
         with:

.gitignore

@@ -72,6 +72,7 @@ yarn-error.log*
 # env files (can opt-in for committing if needed)
 .env*
 !.env.example
+!.env.devin-bridge.example
 !.env.homolog.example
 # Provider API keys (never commit)
 *.api-key
@@ -209,6 +210,8 @@ scripts/i18n/_pending-keys.json
 .agents/
 .antigravitycli/
 .claude/
+!tests/fixtures/devin-bridge/e2e-workspace/.claude/
+!tests/fixtures/devin-bridge/e2e-workspace/.claude/**
 
 # PR Reviews and local feedback files
 pr_reviews*.json
@@ -248,6 +251,8 @@ _artifacts/
 
 # CI/local quality artifacts (eslint-results.json, quality-ratchet.md, etc.)
 .artifacts/
+# Isolated Devin bridge workspaces, evidence, and test databases
+.sandbox/
 
 # Homologation E2E suite (npm run homolog) — real-environment credentials + report output
 .env.homolog

@omniroute/opencode-plugin/README.md

@@ -196,6 +196,7 @@ npm install --prefix ~/.config/opencode/plugins/omniroute-opencode-plugin-prepro
 | 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                   |
+| Model allowlist/blocklist                   | Curate the model picker to a fixed set of IDs via `features.visibleModels` (allowlist) and/or `features.hiddenModels` (blocklist). Bare suffixes like `claude-opus-4-7` match any `{prefix}/claude-opus-4-7`. Both compose with `usableOnly` (all filters AND together). Blocklist wins over allowlist (deny takes precedence)                                                                          | 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     |
@@ -226,6 +227,8 @@ Every field is optional. Defaults mirror v0.1.0 behaviour so existing `opencode.
 | `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` → `GEMINI`). 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.                                                                                                                                                   |
+| `visibleModels`       | `string[]` | _unset_ | Allowlist — when set and non-empty, only models whose raw `/v1/models` ID matches are emitted. Bare IDs (no slash, e.g. `claude-opus-4-7`) match any `{prefix}/claude-opus-4-7`; full IDs (e.g. `cc/claude-opus-4-7`) match exactly. Composes with `usableOnly` and `hiddenModels` (all filters AND together). Unset or empty = no filter.                                                                                                        |
+| `hiddenModels`        | `string[]` | _unset_ | Blocklist — models whose raw ID matches are dropped. Same matching rules as `visibleModels`. When a model is in both `visibleModels` and `hiddenModels`, the blocklist wins (deny takes precedence). Composes with `usableOnly` and `visibleModels` (all filters AND together). Unset or empty = no filter.                                                                                                                                          |
 | `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                                                                                                                                                                                                                                                                                                                                                                                                                               |
@@ -298,7 +301,45 @@ If you want a narrower-scoped Bearer for MCP (different from the chat/inference
 - `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`
+#### Example — curating the model picker (allowlist + blocklist)
+
+A typical OmniRoute instance serves 600+ models. The OpenCode TUI/CLI picker becomes unusable when you need to scroll through hundreds of entries to find the ~30 models you actually use. `visibleModels` and `hiddenModels` let you curate the picker to a fixed set of model IDs that persists in `opencode.json` across config resets.
+
+```jsonc
+{
+  "plugin": [
+    [
+      "@omniroute/opencode-plugin",
+      {
+        "providerId": "omniroute",
+        "baseURL": "https://or.example.com",
+        "features": {
+          "combos": true,
+          "enrichment": true,
+          "usableOnly": true,
+          "visibleModels": [
+            "claude-opus-4-7",     // bare suffix: matches cc/claude-opus-4-7, kr/claude-opus-4-7, etc.
+            "cc/claude-sonnet-4-6", // exact: only the cc/ alias
+            "gemini-2.5-pro",
+            "gpt-5",
+            "o3",
+            "o3-pro",
+            "o4-mini",
+          ],
+          "hiddenModels": [
+            "o3-mini",  // hide the mini variant even if visibleModels is unset
+          ],
+        },
+      },
+    ],
+  ],
+}
+```
+
+- `visibleModels` is an allowlist — only models whose raw ID matches are emitted. Bare IDs (no slash) match any provider prefix; full IDs (with slash) match exactly.
+- `hiddenModels` is a blocklist — listed models are dropped. When a model is in both lists, the blocklist wins (deny takes precedence).
+- Both compose with `usableOnly` (all filters AND together: a model must pass usableOnly AND visibleModels AND not be in hiddenModels).
+- Unset or empty = no filter (current behavior).
 
 [`@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/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@omniroute/opencode-plugin",
-  "version": "0.2.0",
+  "version": "0.2.1",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "@omniroute/opencode-plugin",
-      "version": "0.2.0",
+      "version": "0.2.1",
       "license": "MIT",
       "dependencies": {
         "zod": "^4.4.3"

@omniroute/opencode-plugin/package.json

@@ -23,7 +23,7 @@
   "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/feature-defaults.test.ts tests/usable-combo.test.ts tests/disk-snapshot-perms.test.ts tests/fork-features.test.ts tests/auto-combo-context.test.ts tests/provider-id-routing.test.ts tests/management-read-token.test.ts tests/auto-sync.test.ts",
+    "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/feature-defaults.test.ts tests/usable-combo.test.ts tests/disk-snapshot-perms.test.ts tests/fork-features.test.ts tests/auto-combo-context.test.ts tests/provider-id-routing.test.ts tests/management-read-token.test.ts tests/auto-sync.test.ts tests/model-allowlist.test.ts",
     "prepublishOnly": "npm run clean && npm run build && npm test"
   },
   "keywords": [

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

@@ -177,6 +177,8 @@ const featuresSchema = z
     mcpToken: z.string().min(1).optional(),
     fetchInterceptor: z.boolean().optional(),
     usableOnly: z.boolean().optional(),
+    visibleModels: z.array(z.string().min(1)).optional(),
+    hiddenModels: z.array(z.string().min(1)).optional(),
     diskCache: z.boolean().optional(),
     providerTag: z.boolean().optional(),
     debugLog: z.boolean().optional(),
@@ -241,6 +243,11 @@ export const OMNIROUTE_FEATURE_DEFAULTS = {
   // default-OFF (read sites use `features.X === true`)
   compressionMetadata: false,
   usableOnly: false,
+  // Array flags: unset/empty = no filter. These are not boolean toggles —
+  // they are operator-curated model-ID lists applied in the dynamic and static
+  // hooks alongside usableOnly (all filters AND together).
+  // visibleModels: undefined,  // allowlist — only listed IDs pass
+  // hiddenModels: undefined,  // blocklist — listed IDs are dropped
   mcpAutoEmit: false,
   debugLog: false,
   startupDebug: false,
@@ -330,7 +337,10 @@ function trimLeadingDashes(value: string): string {
  * sees a consistent identifier.
  */
 export function resolveOmniRoutePluginOptions(opts?: OmniRoutePluginOptions): Required<
-  Pick<OmniRoutePluginOptions, "providerId" | "displayName" | "modelCacheTtl" | "autoSyncIntervalMs">
+  Pick<
+    OmniRoutePluginOptions,
+    "providerId" | "displayName" | "modelCacheTtl" | "autoSyncIntervalMs"
+  >
 > & {
   /**
    * #6859: the UNPREFIXED provider id ("omniroute", "omniroute-preprod", …).
@@ -621,7 +631,7 @@ export function createOmniRouteAuthHook(opts?: OmniRoutePluginOptions): AuthHook
  */
 export function invalidateOmniRouteFetchCache(
   cache: OmniRouteFetchCache,
-  baseURL?: string,
+  baseURL?: string
 ): number {
   if (!baseURL) {
     const n = cache.size;
@@ -645,7 +655,7 @@ export function invalidateOmniRouteFetchCache(
  */
 export async function resolveOmniRouteRuntimeAuth(
   resolved: ResolvedOmniRoutePluginOptions,
-  readAuthJson?: OmniRouteReadAuthJson,
+  readAuthJson?: OmniRouteReadAuthJson
 ): Promise<{ apiKey: string; baseURL: string; managementReadToken: string } | null> {
   const reader = readAuthJson ?? defaultReadAuthJson;
   let authJson: AuthJsonShape | undefined | null;
@@ -672,7 +682,7 @@ export async function resolveOmniRouteRuntimeAuth(
       e &&
       (e as { type?: unknown }).type === "api" &&
       typeof (e as { key?: unknown }).key === "string" &&
-      ((e as { key: string }).key).length > 0
+      (e as { key: string }).key.length > 0
     ) {
       entry = e as AuthJsonApiEntry;
       break;
@@ -737,7 +747,7 @@ export async function forceSyncOmniRouteModels(args: {
 
   const auth = await resolveOmniRouteRuntimeAuth(
     resolved,
-    args.readAuthJson ?? defaultReadAuthJson,
+    args.readAuthJson ?? defaultReadAuthJson
   );
   if (!auth) {
     return {
@@ -795,7 +805,7 @@ export async function forceSyncOmniRouteModels(args: {
         rawCompressionCombos = await compressionMetaFetcher(
           auth.baseURL,
           auth.managementReadToken,
-          10_000,
+          10_000
         );
       } catch {
         rawCompressionCombos = [];
@@ -820,10 +830,7 @@ export async function forceSyncOmniRouteModels(args: {
       rawConnections,
       expiresAt: t + resolved.modelCacheTtl,
     };
-    const cacheKey = modelsCacheKey(
-      auth.baseURL,
-      `${auth.apiKey}\0${auth.managementReadToken}`,
-    );
+    const cacheKey = modelsCacheKey(auth.baseURL, `${auth.apiKey}\0${auth.managementReadToken}`);
     cache.set(cacheKey, entry);
 
     if (wantDiskCache) {
@@ -831,7 +838,7 @@ export async function forceSyncOmniRouteModels(args: {
         const fingerprint = diskSnapshotIdentityFingerprint(
           auth.baseURL,
           auth.apiKey,
-          auth.managementReadToken,
+          auth.managementReadToken
         );
         const { expiresAt: _expiresAt, ...diskEntry } = entry;
         await defaultDiskSnapshotWriter(resolved.providerId, diskEntry, fingerprint);
@@ -843,7 +850,7 @@ export async function forceSyncOmniRouteModels(args: {
     console.warn(
       `[omniroute-plugin] force sync ok providerId=${resolved.providerId} ` +
         `models=${rawModels.length} combos=${rawCombos.length} ` +
-        `clearedMemory=${clearedMemory + clearedAll} disk=${clearedDisk}`,
+        `clearedMemory=${clearedMemory + clearedAll} disk=${clearedDisk}`
     );
 
     return {
@@ -944,7 +951,7 @@ export function startOmniRouteAutoSync(args: {
       const result = await forceSyncOmniRouteModels({ resolved, cache });
       if (!result.ok) {
         console.warn(
-          `[omniroute-plugin] auto-sync failed providerId=${resolved.providerId}: ${result.error}`,
+          `[omniroute-plugin] auto-sync failed providerId=${resolved.providerId}: ${result.error}`
         );
         return;
       }
@@ -955,7 +962,7 @@ export function startOmniRouteAutoSync(args: {
       if (result.count !== lastCount) {
         console.warn(
           `[omniroute-plugin] auto-sync catalog size changed ${lastCount} → ${result.count} ` +
-            `(providerId=${resolved.providerId})`,
+            `(providerId=${resolved.providerId})`
         );
         lastCount = result.count;
       }
@@ -976,7 +983,7 @@ export function startOmniRouteAutoSync(args: {
   }
 
   console.warn(
-    `[omniroute-plugin] auto-sync enabled intervalMs=${intervalMs} providerId=${resolved.providerId}`,
+    `[omniroute-plugin] auto-sync enabled intervalMs=${intervalMs} providerId=${resolved.providerId}`
   );
 
   return () => {
@@ -1032,7 +1039,13 @@ export const OmniRoutePlugin: Plugin = async (_input, options) => {
     const cfg = input as Config & {
       command?: Record<
         string,
-        { template: string; description?: string; agent?: string; model?: string; subtask?: boolean }
+        {
+          template: string;
+          description?: string;
+          agent?: string;
+          model?: string;
+          subtask?: boolean;
+        }
       >;
     };
     if (!cfg.command) cfg.command = {};
@@ -2820,6 +2833,118 @@ export function isUsableCombo(
   return false;
 }
 
+// ─────────────────────────────────────────────────────────────────────────
+// #9473 — Model allowlist / blocklist filter helpers
+// ─────────────────────────────────────────────────────────────────────────
+
+/**
+ * Pre-compiled filter structure for the model allowlist/blocklist.
+ *
+ * "exact" holds full raw IDs (e.g. "cc/claude-opus-4-7") for O(1) match.
+ * "suffixes" holds bare model IDs (e.g. "claude-opus-4-7") that match any
+ * "{prefix}/claude-opus-4-7" — so operators can curate by model name without
+ * knowing the provider prefix.
+ */
+export interface ModelListFilter {
+  exact: Set<string>;
+  suffixes: Set<string>;
+}
+
+/**
+ * Compile a string[] of model IDs into a pre-computed filter structure.
+ * Returns undefined when the list is empty or undefined — the "no filter"
+ * state that callers use as a passthrough.
+ *
+ * IDs containing a "/" are stored in "exact"; bare IDs (no slash) go into
+ * "suffixes" and match any "{prefix}/<suffix>" at check time.
+ */
+export function compileModelListFilter(list?: string[]): ModelListFilter | undefined {
+  if (!list || list.length === 0) return undefined;
+  const exact = new Set<string>();
+  const suffixes = new Set<string>();
+  for (const id of list) {
+    if (id.includes("/")) {
+      exact.add(id);
+    } else {
+      suffixes.add(id);
+    }
+  }
+  if (exact.size === 0 && suffixes.size === 0) return undefined;
+  return { exact, suffixes };
+}
+
+/**
+ * Decide whether a raw model ID passes the allowlist/blocklist filter.
+ *
+ * Rules (all filters AND together with usableOnly):
+ *   - No visible filter and no hidden filter → keep (passthrough).
+ *   - Visible filter set: id must match either the exact set or the suffix
+ *     set (bare suffix "claude-opus-4-7" matches any "{prefix}/claude-opus-4-7").
+ *   - Hidden filter set: id must NOT match either the exact or suffix set.
+ *   - If id is in BOTH visible and hidden → DROP (deny wins — safer).
+ *   - No-slash ids (e.g. combo names like "claude-primary") are checked
+ *     against the exact set directly, and against the suffix set as a bare
+ *     match.
+ *
+ * Pure function — exported so static + dynamic hooks share the same
+ * verdict logic without divergence.
+ */
+export function passesModelAllowlist(
+  id: string,
+  visible?: ModelListFilter,
+  hidden?: ModelListFilter
+): boolean {
+  // Hidden filter takes precedence (deny wins over allow).
+  if (hidden) {
+    if (hidden.exact.has(id) || matchesSuffix(id, hidden.suffixes)) return false;
+  }
+  // Visible filter: if set, id must match.
+  if (visible) {
+    if (!visible.exact.has(id) && !matchesSuffix(id, visible.suffixes)) return false;
+  }
+  return true;
+}
+
+/**
+ * Decide whether a combo passes the allowlist filter. A combo keeps when
+ * AT LEAST ONE of its members matches the visible filter. When no visible
+ * filter is set, all combos pass. Combos with zero resolvable members pass
+ * (mirrors `isUsableCombo` semantics).
+ */
+export function passesComboAllowlist(
+  combo: OmniRouteRawCombo,
+  visible?: ModelListFilter
+): boolean {
+  if (!visible) return true;
+  const steps = Array.isArray(combo.models) ? combo.models : [];
+  if (steps.length === 0) return true;
+  let sawResolvableMember = false;
+  for (const step of steps) {
+    if (step?.kind === "combo-ref") continue;
+    const modelId = typeof step?.model === "string" ? step.model : "";
+    if (modelId.length === 0) continue;
+    sawResolvableMember = true;
+    if (visible.exact.has(modelId) || matchesSuffix(modelId, visible.suffixes)) return true;
+  }
+  // No resolvable member → can't prove it should be hidden; keep.
+  if (!sawResolvableMember) return true;
+  // Every resolvable member failed the allowlist → drop.
+  return false;
+}
+
+/**
+ * Check whether a raw model ID matches any suffix in the set.
+ * For an id like `cc/claude-opus-4-7`, the suffix after the first `/`
+ * is checked against the suffixes set. For a bare id like `claude-primary`,
+ * the id itself is checked against the suffixes set.
+ */
+function matchesSuffix(id: string, suffixes: Set<string>): boolean {
+  if (suffixes.size === 0) return false;
+  const slash = id.indexOf("/");
+  const suffix = slash > 0 ? id.slice(slash + 1) : id;
+  return suffixes.has(suffix);
+}
+
 /**
  * Slugify a combo display name into a copy/paste-friendly URL-safe segment.
  * Lowercases, replaces any run of non-alphanumeric chars with a single dash,
@@ -3003,6 +3128,9 @@ export function createOmniRouteProviderHook(
   const wantCompressionMeta = features.compressionMetadata === true;
   const wantUsableOnly = features.usableOnly === true;
   const wantProviderTag = features.providerTag !== false;
+  // #9473: model allowlist/blocklist — compile once per hook instance.
+  const visibleFilter = compileModelListFilter(features.visibleModels);
+  const hiddenFilter = compileModelListFilter(features.hiddenModels);
   const now = deps.now ?? Date.now;
   // T-07: cache holds RAW fetch results (not pre-derived ModelV2) so that
   // the config-shim hook can share the same cache and derive its stripped
@@ -3237,6 +3365,8 @@ export function createOmniRouteProviderHook(
         if (!entry.id) continue;
         if (canonicalDedup.has(entry.id)) continue;
         if (usable && !isUsableRawModelId(entry.id, usable, rawEnrichment)) continue;
+        // #9473: allowlist/blocklist filter (AND with usableOnly).
+        if (!passesModelAllowlist(entry.id, visibleFilter, hiddenFilter)) continue;
         const model = mapRawModelToModelV2(entry, {
           // #6859: server-facing id — NOT the OC-gate-prefixed `resolved.providerId`.
           providerId: resolved.omnirouteProviderId,
@@ -3312,6 +3442,8 @@ export function createOmniRouteProviderHook(
         if (!combo.id) return false;
         if (combo.isHidden === true) return false;
         if (usable && !isUsableCombo(combo, usable)) return false;
+        // #9473: combo allowlist — drop when no member matches visible filter.
+        if (visibleFilter && !passesComboAllowlist(combo, visibleFilter)) return false;
         return true;
       });
       // Resolved nested combos keyed by their friendly name, so parent
@@ -4129,6 +4261,9 @@ export function buildStaticProviderEntry(
     wantUsableOnly && connections && connections.length > 0
       ? usableProviderAliasSet(connections, enrichment)
       : undefined;
+  // #9473: model allowlist/blocklist — compile once per static-block build.
+  const visibleFilter = compileModelListFilter(opts.features?.visibleModels);
+  const hiddenFilter = compileModelListFilter(opts.features?.hiddenModels);
   // Provider-tag suffix — default-on, opt-out via `features.providerTag: false`.
   // Prepends e.g. `Claude - ` to enriched raw-model names so the picker
   // can tell `cc/claude-opus-4-7` (Anthropic) apart from `kr/claude-opus-4-7`
@@ -4166,6 +4301,8 @@ export function buildStaticProviderEntry(
     // Skip canonical-named twins when the alias-keyed enriched row exists.
     if (canonicalDedup.has(raw.id)) continue;
     if (usable && !isUsableRawModelId(raw.id, usable, enrichment)) continue;
+    // #9473: allowlist/blocklist filter (AND with usableOnly).
+    if (!passesModelAllowlist(raw.id, visibleFilter, hiddenFilter)) continue;
     const caps = raw.capabilities ?? {};
     // Enrichment overlay: `/api/pricing/models` carries human display names
     // (e.g. "Claude Opus 4.7" for raw id "cc/claude-opus-4-7"). The OC TUI
@@ -4318,6 +4455,8 @@ export function buildStaticProviderEntry(
     if (!combo.id) return false;
     if (combo.isHidden === true) return false;
     if (usable && !isUsableCombo(combo, usable)) return false;
+    // #9473: combo allowlist — drop when no member matches visible filter.
+    if (visibleFilter && !passesComboAllowlist(combo, visibleFilter)) return false;
     return true;
   });
 
@@ -4466,7 +4605,8 @@ export function buildStaticProviderEntry(
       // (`opencode-omniroute/opencode-omniroute/<slug>`), and `parseModel()`
       // resolves credentials for the nonexistent provider `opencode-omniroute`
       // instead of `omniroute`. See #7976.
-      models[buildComboKey(combo, usedComboKeys, opts.omnirouteProviderId).split("/").pop()!] = entry;
+      models[buildComboKey(combo, usedComboKeys, opts.omnirouteProviderId).split("/").pop()!] =
+        entry;
 
       // Make this combo's resolved entry available to parent combos
       // that reference it via combo-ref. Use the friendly name since

@omniroute/opencode-plugin/tests/model-allowlist.test.ts

@@ -0,0 +1,317 @@
+/**
+ * #9473 — Model allowlist/blocklist for the opencode-plugin.
+ *
+ * Tests for the pure filter helpers (`compileModelListFilter`,
+ * `passesModelAllowlist`, `passesComboAllowlist`) and the schema + hook-level
+ * integration. The allowlist/blocklist composes with `usableOnly` (all filters
+ * AND together), blocklist wins over allowlist (deny takes precedence), and
+ * bare-suffix entries (e.g. "claude-opus-4-7") match any "{prefix}/claude-opus-4-7".
+ */
+
+import test from "node:test";
+import assert from "node:assert/strict";
+
+import {
+  compileModelListFilter,
+  passesModelAllowlist,
+  passesComboAllowlist,
+  parseOmniRoutePluginOptions,
+  buildStaticProviderEntry,
+  resolveOmniRoutePluginOptions,
+  type OmniRouteRawCombo,
+  type OmniRouteRawModelEntry,
+} from "../src/index.js";
+
+// ─────────────────────────────────────────────────────────────────────────
+// compileModelListFilter
+// ─────────────────────────────────────────────────────────────────────────
+
+test("compileModelListFilter: undefined list → undefined", () => {
+  assert.equal(compileModelListFilter(undefined), undefined);
+});
+
+test("compileModelListFilter: empty array → undefined", () => {
+  assert.equal(compileModelListFilter([]), undefined);
+});
+
+test("compileModelListFilter: raw IDs with slash → exact set populated", () => {
+  const f = compileModelListFilter(["cc/claude-opus-4-7", "glm/gpt-5"]);
+  assert.ok(f);
+  assert.equal(f.exact.has("cc/claude-opus-4-7"), true);
+  assert.equal(f.exact.has("glm/gpt-5"), true);
+  assert.equal(f.suffixes.size, 0);
+});
+
+test("compileModelListFilter: bare IDs (no slash) → suffixes set populated", () => {
+  const f = compileModelListFilter(["claude-opus-4-7", "gpt-5"]);
+  assert.ok(f);
+  assert.equal(f.suffixes.has("claude-opus-4-7"), true);
+  assert.equal(f.suffixes.has("gpt-5"), true);
+  assert.equal(f.exact.size, 0);
+});
+
+test("compileModelListFilter: mixed raw + bare → both sets populated", () => {
+  const f = compileModelListFilter(["cc/claude-opus-4-7", "gpt-5"]);
+  assert.ok(f);
+  assert.equal(f.exact.has("cc/claude-opus-4-7"), true);
+  assert.equal(f.suffixes.has("gpt-5"), true);
+});
+
+// ─────────────────────────────────────────────────────────────────────────
+// passesModelAllowlist
+// ─────────────────────────────────────────────────────────────────────────
+
+test("passesModelAllowlist: no visible, no hidden → keep (passthrough)", () => {
+  assert.equal(passesModelAllowlist("cc/claude-opus-4-7", undefined, undefined), true);
+});
+
+test("passesModelAllowlist: visible undefined, hidden undefined → keep", () => {
+  assert.equal(passesModelAllowlist("cc/claude-opus-4-7", undefined, undefined), true);
+});
+
+test("passesModelAllowlist: visible set, id matches exact → keep", () => {
+  const vis = compileModelListFilter(["cc/claude-opus-4-7"]);
+  assert.equal(passesModelAllowlist("cc/claude-opus-4-7", vis, undefined), true);
+});
+
+test("passesModelAllowlist: visible set, id matches suffix → keep", () => {
+  const vis = compileModelListFilter(["claude-opus-4-7"]);
+  assert.equal(passesModelAllowlist("cc/claude-opus-4-7", vis, undefined), true);
+});
+
+test("passesModelAllowlist: visible set, id does NOT match → drop", () => {
+  const vis = compileModelListFilter(["cc/claude-opus-4-7"]);
+  assert.equal(passesModelAllowlist("glm/gpt-5", vis, undefined), false);
+});
+
+test("passesModelAllowlist: visible set, bare suffix matches different prefix → keep", () => {
+  const vis = compileModelListFilter(["claude-opus-4-7"]);
+  assert.equal(passesModelAllowlist("kr/claude-opus-4-7", vis, undefined), true);
+});
+
+test("passesModelAllowlist: hidden set, id matches exact → drop", () => {
+  const hid = compileModelListFilter(["cc/claude-opus-4-7"]);
+  assert.equal(passesModelAllowlist("cc/claude-opus-4-7", undefined, hid), false);
+});
+
+test("passesModelAllowlist: hidden set, id matches suffix → drop", () => {
+  const hid = compileModelListFilter(["claude-opus-4-7"]);
+  assert.equal(passesModelAllowlist("cc/claude-opus-4-7", undefined, hid), false);
+});
+
+test("passesModelAllowlist: hidden set, id does NOT match → keep", () => {
+  const hid = compileModelListFilter(["cc/claude-opus-4-7"]);
+  assert.equal(passesModelAllowlist("glm/gpt-5", undefined, hid), true);
+});
+
+test("passesModelAllowlist: id in BOTH visible and hidden → DROP (deny wins)", () => {
+  const vis = compileModelListFilter(["cc/claude-opus-4-7"]);
+  const hid = compileModelListFilter(["cc/claude-opus-4-7"]);
+  assert.equal(passesModelAllowlist("cc/claude-opus-4-7", vis, hid), false);
+});
+
+test("passesModelAllowlist: visible allows, hidden blocks different id → keep the visible one", () => {
+  const vis = compileModelListFilter(["cc/claude-opus-4-7"]);
+  const hid = compileModelListFilter(["glm/gpt-5"]);
+  assert.equal(passesModelAllowlist("cc/claude-opus-4-7", vis, hid), true);
+  assert.equal(passesModelAllowlist("glm/gpt-5", vis, hid), false);
+});
+
+test("passesModelAllowlist: bare-suffix hidden blocks exact match too", () => {
+  const hid = compileModelListFilter(["claude-opus-4-7"]);
+  assert.equal(passesModelAllowlist("cc/claude-opus-4-7", undefined, hid), false);
+  assert.equal(passesModelAllowlist("kr/claude-opus-4-7", undefined, hid), false);
+});
+
+test("passesModelAllowlist: no-slash id, visible set has bare match → keep", () => {
+  const vis = compileModelListFilter(["claude-primary"]);
+  assert.equal(passesModelAllowlist("claude-primary", vis, undefined), true);
+});
+
+test("passesModelAllowlist: no-slash id, visible set has no match → drop", () => {
+  const vis = compileModelListFilter(["cc/claude-opus-4-7"]);
+  assert.equal(passesModelAllowlist("claude-primary", vis, undefined), false);
+});
+
+// ─────────────────────────────────────────────────────────────────────────
+// passesComboAllowlist
+// ─────────────────────────────────────────────────────────────────────────
+
+function combo(models: OmniRouteRawCombo["models"]): OmniRouteRawCombo {
+  return { id: "c1", name: "Test Combo", models };
+}
+
+test("passesComboAllowlist: visible undefined → keep", () => {
+  const c = combo([{ kind: "model", model: "cc/claude-opus-4-7" }]);
+  assert.equal(passesComboAllowlist(c, undefined), true);
+});
+
+test("passesComboAllowlist: ≥1 member matches visible → keep", () => {
+  const vis = compileModelListFilter(["cc/claude-opus-4-7"]);
+  const c = combo([
+    { kind: "model", model: "dead/legacy" },
+    { kind: "model", model: "cc/claude-opus-4-7" },
+  ]);
+  assert.equal(passesComboAllowlist(c, vis), true);
+});
+
+test("passesComboAllowlist: zero members match visible → drop", () => {
+  const vis = compileModelListFilter(["cc/claude-opus-4-7"]);
+  const c = combo([
+    { kind: "model", model: "glm/gpt-5" },
+    { kind: "model", model: "kr/claude-opus-4-7" },
+  ]);
+  assert.equal(passesComboAllowlist(c, vis), false);
+});
+
+test("passesComboAllowlist: bare suffix matches any prefix → keep", () => {
+  const vis = compileModelListFilter(["claude-opus-4-7"]);
+  const c = combo([{ kind: "model", model: "kr/claude-opus-4-7" }]);
+  assert.equal(passesComboAllowlist(c, vis), true);
+});
+
+test("passesComboAllowlist: zero members → keep", () => {
+  const vis = compileModelListFilter(["cc/claude-opus-4-7"]);
+  assert.equal(passesComboAllowlist(combo([]), vis), true);
+  assert.equal(passesComboAllowlist(combo(undefined), vis), true);
+});
+
+test("passesComboAllowlist: only combo-ref steps → keep", () => {
+  const vis = compileModelListFilter(["cc/claude-opus-4-7"]);
+  const c = combo([{ kind: "combo-ref", comboName: "nested" }]);
+  assert.equal(passesComboAllowlist(c, vis), true);
+});
+
+// ─────────────────────────────────────────────────────────────────────────
+// Schema — visibleModels / hiddenModels
+// ─────────────────────────────────────────────────────────────────────────
+
+test("parseOmniRoutePluginOptions: visibleModels string[] → preserved", () => {
+  const r = parseOmniRoutePluginOptions({
+    features: { visibleModels: ["cc/claude-opus-4-7", "gpt-5"] },
+  });
+  assert.deepEqual(r.features?.visibleModels, ["cc/claude-opus-4-7", "gpt-5"]);
+});
+
+test("parseOmniRoutePluginOptions: hiddenModels string[] → preserved", () => {
+  const r = parseOmniRoutePluginOptions({
+    features: { hiddenModels: ["glm/gpt-5"] },
+  });
+  assert.deepEqual(r.features?.hiddenModels, ["glm/gpt-5"]);
+});
+
+test("parseOmniRoutePluginOptions: both lists together → preserved", () => {
+  const r = parseOmniRoutePluginOptions({
+    features: {
+      visibleModels: ["cc/claude-opus-4-7"],
+      hiddenModels: ["glm/gpt-5"],
+    },
+  });
+  assert.deepEqual(r.features?.visibleModels, ["cc/claude-opus-4-7"]);
+  assert.deepEqual(r.features?.hiddenModels, ["glm/gpt-5"]);
+});
+
+test("parseOmniRoutePluginOptions: empty string in visibleModels → rejects", () => {
+  assert.throws(
+    () =>
+      parseOmniRoutePluginOptions({
+        features: { visibleModels: [""] },
+      }),
+    /Invalid @omniroute\/opencode-plugin options/
+  );
+});
+
+test("parseOmniRoutePluginOptions: empty string in hiddenModels → rejects", () => {
+  assert.throws(
+    () =>
+      parseOmniRoutePluginOptions({
+        features: { hiddenModels: [""] },
+      }),
+    /Invalid @omniroute\/opencode-plugin options/
+  );
+});
+
+test("parseOmniRoutePluginOptions: unknown features key still rejects (strict invariant)", () => {
+  assert.throws(
+    () =>
+      parseOmniRoutePluginOptions({
+        features: { visibleModels: ["x"], unknownKey: true },
+      }),
+    /Invalid @omniroute\/opencode-plugin options/
+  );
+});
+
+// ─────────────────────────────────────────────────────────────────────────
+// buildStaticProviderEntry — allowlist/blocklist integration
+// ─────────────────────────────────────────────────────────────────────────
+
+const FAKE_RAW_MODELS: OmniRouteRawModelEntry[] = [
+  { id: "cc/claude-opus-4-7", owned_by: "anthropic" },
+  { id: "glm/gpt-5", owned_by: "openai" },
+  { id: "kr/claude-opus-4-7", owned_by: "anthropic" },
+  { id: "claude-primary", owned_by: "combo" },
+];
+
+test("buildStaticProviderEntry: no allowlist → all models emitted", () => {
+  const opts = resolveOmniRoutePluginOptions({ features: {} });
+  const entry = buildStaticProviderEntry(FAKE_RAW_MODELS, [], opts, "http://localhost:20128/v1", "sk-test");
+  const ids = Object.keys(entry.models);
+  assert.ok(ids.includes("cc/claude-opus-4-7"), "cc/claude-opus-4-7 should be present");
+  assert.ok(ids.includes("glm/gpt-5"), "glm/gpt-5 should be present");
+  assert.ok(ids.includes("kr/claude-opus-4-7"), "kr/claude-opus-4-7 should be present");
+});
+
+test("buildStaticProviderEntry: visibleModels filters to only listed IDs", () => {
+  const opts = resolveOmniRoutePluginOptions({
+    features: { visibleModels: ["cc/claude-opus-4-7"] },
+  });
+  const entry = buildStaticProviderEntry(FAKE_RAW_MODELS, [], opts, "http://localhost:20128/v1", "sk-test");
+  const ids = Object.keys(entry.models);
+  assert.ok(ids.includes("cc/claude-opus-4-7"), "cc/claude-opus-4-7 should be present");
+  assert.equal(ids.includes("glm/gpt-5"), false, "glm/gpt-5 should be filtered out");
+  assert.equal(ids.includes("kr/claude-opus-4-7"), false, "kr/claude-opus-4-7 should be filtered out");
+});
+
+test("buildStaticProviderEntry: hiddenModels drops listed IDs", () => {
+  const opts = resolveOmniRoutePluginOptions({
+    features: { hiddenModels: ["glm/gpt-5"] },
+  });
+  const entry = buildStaticProviderEntry(FAKE_RAW_MODELS, [], opts, "http://localhost:20128/v1", "sk-test");
+  const ids = Object.keys(entry.models);
+  assert.ok(ids.includes("cc/claude-opus-4-7"), "cc/claude-opus-4-7 should be present");
+  assert.equal(ids.includes("glm/gpt-5"), false, "glm/gpt-5 should be hidden");
+  assert.ok(ids.includes("kr/claude-opus-4-7"), "kr/claude-opus-4-7 should be present");
+});
+
+test("buildStaticProviderEntry: bare-suffix visibleModels matches any prefix", () => {
+  const opts = resolveOmniRoutePluginOptions({
+    features: { visibleModels: ["claude-opus-4-7"] },
+  });
+  const entry = buildStaticProviderEntry(FAKE_RAW_MODELS, [], opts, "http://localhost:20128/v1", "sk-test");
+  const ids = Object.keys(entry.models);
+  assert.ok(ids.includes("cc/claude-opus-4-7"), "cc/claude-opus-4-7 should match via suffix");
+  assert.ok(ids.includes("kr/claude-opus-4-7"), "kr/claude-opus-4-7 should match via suffix");
+  assert.equal(ids.includes("glm/gpt-5"), false, "glm/gpt-5 should be filtered out");
+});
+
+test("buildStaticProviderEntry: id in both visible and hidden → hidden wins", () => {
+  const opts = resolveOmniRoutePluginOptions({
+    features: {
+      visibleModels: ["cc/claude-opus-4-7"],
+      hiddenModels: ["cc/claude-opus-4-7"],
+    },
+  });
+  const entry = buildStaticProviderEntry(FAKE_RAW_MODELS, [], opts, "http://localhost:20128/v1", "sk-test");
+  const ids = Object.keys(entry.models);
+  assert.equal(ids.includes("cc/claude-opus-4-7"), false, "deny takes precedence");
+});
+
+test("buildStaticProviderEntry: empty visibleModels → no filter (passthrough)", () => {
+  const opts = resolveOmniRoutePluginOptions({
+    features: { visibleModels: [] },
+  });
+  const entry = buildStaticProviderEntry(FAKE_RAW_MODELS, [], opts, "http://localhost:20128/v1", "sk-test");
+  const ids = Object.keys(entry.models);
+  assert.ok(ids.includes("cc/claude-opus-4-7"), "empty visibleModels should not filter");
+  assert.ok(ids.includes("glm/gpt-5"), "empty visibleModels should not filter");
+});

AGENTS.md

@@ -58,7 +58,7 @@ Repository map and Reference Documentation sections below.
 | Services      | `open-sse/services/`    | Combo routing, rate limits, caching, etc                                                                                                                |
 | Database      | `src/lib/db/`           | SQLite domain modules (130 migrations)                                                                                                                  |
 | Domain/Policy | `src/domain/`           | Policy engine, cost rules, fallback logic                                                                                                               |
-| MCP Server    | `open-sse/mcp-server/`  | 104 tools (42 base + memory/skill/agentSkill/pool/notion/obsidian/gamification/plugin modules), 3 transports (stdio / SSE / Streamable HTTP), 31 scopes |
+| MCP Server    | `open-sse/mcp-server/`  | 105 tools (42 base + memory/skill/agentSkill/pool/notion/obsidian/gamification/plugin modules), 3 transports (stdio / SSE / Streamable HTTP), 31 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                                                                                                                        |
@@ -399,6 +399,7 @@ For any non-trivial change, read the matching deep-dive first:
 | Resilience (3 mechanisms)                     | `docs/architecture/RESILIENCE_GUIDE.md`                 |
 | Reasoning replay                              | `docs/routing/REASONING_REPLAY.md`                      |
 | Skills framework                              | `docs/frameworks/SKILLS.md`                             |
+| Radar (free-model catalog overlay)            | `docs/frameworks/RADAR.md`                              |
 | Memory system (FTS5 + Qdrant)                 | `docs/frameworks/MEMORY.md`                             |
 | Cloud agents                                  | `docs/frameworks/CLOUD_AGENT.md`                        |
 | Guardrails (PII / injection / vision)         | `docs/security/GUARDRAILS.md`                           |
@@ -427,7 +428,7 @@ For any non-trivial change, read the matching deep-dive first:
 | What                    | Command                                                                     |
 | ----------------------- | --------------------------------------------------------------------------- |
 | Unit tests              | `npm run test:unit`                                                         |
-| Single file             | `node --import tsx/esm --test tests/unit/your-file.test.ts`                      |
+| Single file             | `node --import tsx/esm --test tests/unit/your-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`                                                |

CONTRIBUTING.md

@@ -15,6 +15,12 @@ coverage, and reconciliation steps.
 
 - **Node.js** `>=22.22.3 <23`, or `>=24.0.0 <27` (recommended: 24 LTS)
 - **npm** 10+
+
+> **npm v11+ users (Node 24+):** After `npm install`, verify native modules were installed:
+> `node -e "require('better-sqlite3')"`. If it fails with `MODULE_NOT_FOUND`,
+> run `npm approve-scripts better-sqlite3 && npm install`. See
+> [Troubleshooting](docs/guides/TROUBLESHOOTING.md#npm-v11-better-sqlite3-not-installed-cannot-find-module).
+
 - **Git**
 
 ### Clone & Install

Dockerfile

@@ -77,7 +77,7 @@ RUN test -f package-lock.json \
 # a broken/rate-limited fetch fails the BUILD loudly instead of shipping a
 # broken image.
 RUN --mount=type=cache,id=npm-cache,target=/root/.npm \
-  npm ci --no-audit --no-fund --legacy-peer-deps --ignore-scripts \
+  npm ci --include=optional --no-audit --no-fund --legacy-peer-deps --ignore-scripts \
   && (cd node_modules/better-sqlite3 \
       && node /usr/local/lib/node_modules/npm/node_modules/node-gyp/bin/node-gyp.js rebuild) \
   && node -e "require('better-sqlite3')(':memory:').close()" \
@@ -119,7 +119,9 @@ ENV NODE_OPTIONS="--max-old-space-size=${OMNIROUTE_BUILD_MEMORY_MB}"
 
 COPY . ./
 RUN --mount=type=cache,id=next-cache,target=/app/.build/next/cache \
-  mkdir -p /app/data && npm run build
+  mkdir -p /app/data \
+  && npm run build \
+  && node --input-type=module -e "import { createRequire } from 'node:module'; import { pathToFileURL } from 'node:url'; const standaloneRoot = '/app/.build/next/standalone/node_modules/'; const require = createRequire('/app/.build/next/standalone/package.json'); for (const pkg of ['@atjsh/llmlingua-2', '@huggingface/transformers', '@tensorflow/tfjs', 'js-tiktoken']) { const resolved = require.resolve(pkg); if (!resolved.startsWith(standaloneRoot)) throw new Error(pkg + ' resolved outside standalone: ' + resolved); await import(pathToFileURL(resolved).href); } const onnxRuntime = require.resolve('onnxruntime-node'); if (!onnxRuntime.startsWith(standaloneRoot)) throw new Error('onnxruntime-node resolved outside standalone: ' + onnxRuntime); await import(pathToFileURL(onnxRuntime).href);"
 
 # ── Runner base ────────────────────────────────────────────────────────────
 FROM base AS runner-base

README.md

@@ -188,7 +188,7 @@ curl http://localhost:20128/v1/chat/completions \
 
 </div>
 
-<img src="./docs/diagrams/promise-pillars.svg" width="100%" alt="The Promise — One endpoint. 291 providers. Never stop building — OmniRoute picks the cheapest one that works. Six pillars: Never hit limits (auto-fallback across 291 providers in milliseconds, zero downtime) · Save up to 95% tokens (RTK + Caveman stacked compression cuts 15–95%, ~89% avg on tool-heavy sessions) · $0 to start (90+ free tiers, 40+ free forever — no card needed) · Every tool works (33 coding agents through one config) · One endpoint (OpenAI ↔ Claude ↔ Gemini ↔ Responses API at /v1) · Production-grade (circuit breakers, TLS stealth, MCP 104 tools, A2A, memory, guardrails, evals — 25,000+ tests)."/>
+<img src="./docs/diagrams/promise-pillars.svg" width="100%" alt="The Promise — One endpoint. 291 providers. Never stop building — OmniRoute picks the cheapest one that works. Six pillars: Never hit limits (auto-fallback across 291 providers in milliseconds, zero downtime) · Save up to 95% tokens (RTK + Caveman stacked compression cuts 15–95%, ~89% avg on tool-heavy sessions) · $0 to start (90+ free tiers, 40+ free forever — no card needed) · Every tool works (33 coding agents through one config) · One endpoint (OpenAI ↔ Claude ↔ Gemini ↔ Responses API at /v1) · Production-grade (circuit breakers, TLS stealth, MCP 105 tools, A2A, memory, guardrails, evals — 25,000+ tests)."/>
 
 <br/>
 <br/>
@@ -439,7 +439,7 @@ All **19** strategies — mix & match per combo step:
 
 </div>
 
-<img src="./docs/diagrams/comparison-table.svg" width="100%" alt="What sets OmniRoute apart — comparison table vs 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute: 291 providers, 90+ free providers built-in, 19 routing strategies, 12-engine token compression, built-in MCP server with 104 tools, A2A agent protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, Desktop/Termux/PWA, 43 i18n UI locales, 100% MIT self-hosted. OmniRoute is the only one with the full set; competitors show a mix of checks, partials and crosses. Verified from each project&apos;s docs."/>
+<img src="./docs/diagrams/comparison-table.svg" width="100%" alt="What sets OmniRoute apart — comparison table vs 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute: 291 providers, 90+ free providers built-in, 19 routing strategies, 12-engine token compression, built-in MCP server with 105 tools, A2A agent protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, Desktop/Termux/PWA, 43 i18n UI locales, 100% MIT self-hosted. OmniRoute is the only one with the full set; competitors show a mix of checks, partials and crosses. Verified from each project&apos;s docs."/>
 
 <sub>📊 Full methodology &amp; per-feature detail vs 9router, OpenRouter, CLIProxyAPI &amp; LiteLLM → [`docs/comparison/OMNIROUTE_VS_ALTERNATIVES.md`](docs/comparison/OMNIROUTE_VS_ALTERNATIVES.md)</sub>
 
@@ -723,7 +723,7 @@ Expose OmniRoute over **MCP**, **A2A**, a **REST API**, **webhooks** or a **remo
 <table>
   <tr><th align="left">Interface</th><th align="left">Endpoint / command</th><th align="left">Use it for</th></tr>
   <tr><td align="left" nowrap>🧰 <b>MCP (stdio)</b></td><td align="left" nowrap><code>omniroute --mcp</code></td><td align="left">Plug into Claude Desktop, Cursor, any MCP client</td></tr>
-  <tr><td align="left" nowrap>🌊 <b>MCP (HTTP)</b></td><td align="left" nowrap><code>/api/mcp/stream</code></td><td align="left">Remote MCP — <b>104 tools</b>, 31 scopes, full audit trail</td></tr>
+  <tr><td align="left" nowrap>🌊 <b>MCP (HTTP)</b></td><td align="left" nowrap><code>/api/mcp/stream</code></td><td align="left">Remote MCP — <b>105 tools</b>, 31 scopes, full audit trail</td></tr>
   <tr><td align="left" nowrap>📡 <b>MCP (SSE)</b></td><td align="left" nowrap><code>/api/mcp/sse</code></td><td align="left">Streaming MCP transport</td></tr>
   <tr><td align="left" nowrap>🤝 <b>A2A</b></td><td align="left" nowrap><code>/.well-known/agent.json</code></td><td align="left">Agent-to-agent, <b>JSON-RPC 2.0</b> + SSE, 6 skills</td></tr>
   <tr><td align="left" nowrap>🌐 <b>REST API</b></td><td align="left" nowrap><code>/v1/*</code></td><td align="left">OpenAI-compatible — chat, embeddings, images, audio, OCR</td></tr>
@@ -890,6 +890,12 @@ docker run -d --name omniroute --restart unless-stopped --stop-timeout 40 \
   -p 127.0.0.1:20128:20128 -v omniroute-data:/app/data diegosouzapw/omniroute:latest
 ```
 
+> **Pre-release Docker channel:** `diegosouzapw/omniroute:next` and
+> `diegosouzapw/omniroute:next-web` follow the current default `release/v*`
+> branch. These mutable tags are intended only for testing unreleased fixes and
+> are **not supported for production**. See
+> [Docker Release Channels](docs/guides/DOCKER_RELEASE_CHANNELS.md).
+
 **🛠️ From source**
 
 ```bash

bin/cli/commands/doctor.mjs

@@ -299,7 +299,15 @@ async function checkNativeBinary(rootDir) {
       "Release",
       "better_sqlite3.node"
     ),
-    path.join(rootDir, "dist", "node_modules", "better-sqlite3", "build", "Release", "better_sqlite3.node"),
+    path.join(
+      rootDir,
+      "dist",
+      "node_modules",
+      "better-sqlite3",
+      "build",
+      "Release",
+      "better_sqlite3.node"
+    ),
     path.join(rootDir, "node_modules", "better-sqlite3", "build", "Release", "better_sqlite3.node"),
   ];
   const binaryPath = candidates.find((candidate) => fs.existsSync(candidate));
@@ -396,7 +404,10 @@ async function checkServerLiveness(options = {}) {
   // First attempt: configured health endpoint (may require auth token).
   const primary = await probeUrl(url);
   if (primary.ok) {
-    return ok("Server liveness", "Server health endpoint is reachable", { url, status: primary.status });
+    return ok("Server liveness", "Server health endpoint is reachable", {
+      url,
+      status: primary.status,
+    });
   }
 
   // #6162: /api/health and /api/health/degradation require a management token.
@@ -427,7 +438,12 @@ async function checkServerLiveness(options = {}) {
     return ok(
       "Server liveness",
       `Server reachable (health endpoint returned ${primary.status}, likely requires MANAGEMENT_TOKEN)`,
-      { primaryUrl: url, primaryStatus: primary.status, fallbackUrl, fallbackStatus: fallback.status }
+      {
+        primaryUrl: url,
+        primaryStatus: primary.status,
+        fallbackUrl,
+        fallbackStatus: fallback.status,
+      }
     );
   }
 
@@ -440,8 +456,7 @@ async function checkServerLiveness(options = {}) {
 
 export async function collectDoctorChecks(context = {}, options = {}) {
   const rootDir =
-    context.rootDir ||
-    path.resolve(path.dirname(fileURLToPath(import.meta.url)), "..", "..", "..");
+    context.rootDir || path.resolve(path.dirname(fileURLToPath(import.meta.url)), "..", "..", "..");
   const dataDir = resolveDataDir();
   const dbPath = resolveStoragePath(dataDir);
 

bin/cli/commands/oauth.mjs

@@ -159,9 +159,7 @@ async function runBrowserFlow(def, opts) {
   }
   const result = await exchangeRes.json();
   const conn = result.connection ?? {};
-  process.stdout.write(
-    `Authorized: ${conn.email ?? conn.displayName ?? conn.id ?? "connected"}\n`
-  );
+  process.stdout.write(`Authorized: ${conn.email ?? conn.displayName ?? conn.id ?? "connected"}\n`);
 }
 
 async function safeErrorBody(res) {

bin/cli/commands/setup-claude.mjs

@@ -160,8 +160,7 @@ export async function runSetupClaudeCommand(opts = {}) {
       let detail = `HTTP ${res.status}`;
       try {
         const errorBody = await res.json();
-        const serverMsg =
-          errorBody?.error?.message || errorBody?.error || errorBody?.message || "";
+        const serverMsg = errorBody?.error?.message || errorBody?.error || errorBody?.message || "";
         if (serverMsg) detail += ` — ${serverMsg}`;
       } catch {}
       throw new Error(detail);

bin/cli/commands/update.mjs

@@ -188,16 +188,18 @@ export async function runUpdateCommand(opts = {}) {
     const afterVersion = await getCurrentVersion();
     if (afterVersion && compareVersions(afterVersion, latest) < 0) {
       printError(
-        `Global install updated to ${latest}, but the running binary still reports ${afterVersion}.`,
+        `Global install updated to ${latest}, but the running binary still reports ${afterVersion}.`
       );
       console.log(
-        "  A local `node_modules/omniroute` is likely shadowing the global install on PATH.",
+        "  A local `node_modules/omniroute` is likely shadowing the global install on PATH."
       );
       console.log("  Diagnose with:");
       console.log("    which -a omniroute");
       console.log("    command -v omniroute");
       console.log("    npm prefix -g");
-      console.log("  Then remove the shadowing local copy (e.g. `npm uninstall omniroute` from its directory)");
+      console.log(
+        "  Then remove the shadowing local copy (e.g. `npm uninstall omniroute` from its directory)"
+      );
       console.log("  or reorder PATH so the global bin comes first.");
       return 1;
     }

bin/cli/runtime/nativeDeps.mjs

@@ -94,10 +94,12 @@ export function isBetterSqliteBinaryValid() {
     const magic = buf.toString("hex");
     const os = platform();
     let formatOk;
-    if (os === "linux") formatOk = magic.startsWith("7f454c46"); // ELF
+    if (os === "linux")
+      formatOk = magic.startsWith("7f454c46"); // ELF
     else if (os === "darwin")
       formatOk = magic.startsWith("cffaedfe") || magic.startsWith("cefaedfe"); // Mach-O
-    else if (os === "win32") formatOk = magic.startsWith("4d5a"); // PE/MZ
+    else if (os === "win32")
+      formatOk = magic.startsWith("4d5a"); // PE/MZ
     else formatOk = true;
     if (!formatOk) return false;
     // File-format magic bytes alone do not guarantee the binary was built for the Node ABI

changelog.d/features/8237-reduce-idle-memory-small-vps.md

@@ -0,0 +1 @@
+- **docs:** add low-memory/small VPS optimization guide ([#8237](https://github.com/diegosouzapw/OmniRoute/issues/8237))

changelog.d/features/8369-connection-level-upstream-headers.md

@@ -0,0 +1 @@
+- **feat(providers):** add connection-level custom upstream headers via `provider_specific_data.customHeaders` — applied to every request through that connection, with model-level headers overriding on the same case-insensitive name. (thanks @Benson-mk)

changelog.d/features/8461-runomniroutecli-approval-gate.md

@@ -0,0 +1 @@
+- feat(copilot): add approval gate for runOmniRouteCli commands (#8461)

changelog.d/features/8799-electron-remote-server-mode.md

@@ -0,0 +1 @@
+- **feat(electron):** Desktop app can now attach to an already-running OmniRoute server (e.g. a Docker/OrbStack container) instead of always spawning its own bundled server — configurable via the tray's "Remote Server → Connect to Remote Server…" or the `OMNIROUTE_REMOTE_URL` env var ([#8799](https://github.com/diegosouzapw/OmniRoute/pull/8799)) — thanks @soulhakr

changelog.d/features/9031-regolo-ai-provider.md

@@ -0,0 +1 @@
+- **feat(providers):** add Regolo AI OpenAI-compatible provider ([#9031](https://github.com/diegosouzapw/OmniRoute/issues/9031))

changelog.d/features/9068-editable-discovered-provider-slugs.md

@@ -0,0 +1 @@
+- **feat(db):** add provider-scoped model aliases that survive rediscovery ([#9068](https://github.com/diegosouzapw/OmniRoute/issues/9068))

changelog.d/features/9243-forwarded-header-budget-env.md

@@ -0,0 +1 @@
+- feat: make forwarded upstream response-header budget configurable via env var (#9243)

changelog.d/features/9415-newapi-sub2api-aggregator-balance.md

@@ -0,0 +1 @@
+- **sse:** New-API / One-API / Sub2API aggregator balance detection for compatible nodes — with the "Aggregator Gateway" toggle on, OmniRoute queries the aggregator's `/api/user/self` to read the account balance, shows it as a dashboard badge and lets quota-preflight routing skip exhausted accounts. Gated by the `NEWAPI_AGGREGATOR_BALANCE` feature flag (default off), with a `quotaPerUnit` override for aggregators that do not use the default 500000 units/$1 rate ([#9415](https://github.com/diegosouzapw/OmniRoute/issues/9415))

changelog.d/features/9418-disable-auto-no-think-combos.md

@@ -0,0 +1 @@
+- **feat(catalog):** added opt-in settings `hideAutoCombos` and `hideNoThinkVariants` (default off) to filter built-in `auto/*` virtual combos and `no-think/*` gateway variants from the `/v1/models` catalog — user-defined combos and original provider models stay listed; routing is unaffected ([#9418](https://github.com/diegosouzapw/OmniRoute/issues/9418))

changelog.d/features/9473-opencode-plugin-model-allowlist.md

@@ -0,0 +1 @@
+- **feat(opencode-plugin):** added `features.visibleModels` (allowlist) and `features.hiddenModels` (blocklist) to `@omniroute/opencode-plugin` — curate the OpenCode TUI/CLI model picker from 600+ catalog entries down to an operator-defined ID list that persists in `opencode.json` across config resets ([#9473](https://github.com/diegosouzapw/OmniRoute/issues/9473))

changelog.d/features/9511-quota-exhausted-transient-test-all.md

@@ -0,0 +1 @@
+- **feat(models):** Test All's "Auto-hide failed models" no longer hides quota errors — daily-quota-exhausted and credits-exhausted responses are now classified via the routing path's existing quota detectors, so an evening Test All on a free-tier provider no longer silently wipes the catalog. Quota results stay visible with a distinct amber badge ([#9511](https://github.com/diegosouzapw/OmniRoute/issues/9511))

changelog.d/features/9570-plugin-context-headers.md

@@ -0,0 +1 @@
+- feat(plugins): expose client request headers in plugin onRequest/onResponse context (#9570)

changelog.d/fixes/8739-yuanbao-sse-parser.md

@@ -0,0 +1 @@
+- fix(yuanbao-web): accept `content` field in SSE text events (upstream format change) (#8739)

changelog.d/fixes/8813-chatgpt-sentinel.md

@@ -0,0 +1 @@
+- fix(errorClassifier): classify ChatGPT Web SENTINEL_BLOCKED 403 as terminal FORBIDDEN, enabling proper combo fallback (#8813)

changelog.d/fixes/8946-no-tool-output.md

@@ -0,0 +1 @@
+- fix(compression): drop orphan custom_tool_call/local_shell_call/apply_patch_call on compaction restore (#8946)

changelog.d/fixes/8994-vertex-partner-claude.md

@@ -0,0 +1 @@
+- fix(vertex): route Claude models to native rawPredict endpoint and respect custom targetFormat overrides (#8994)

changelog.d/fixes/9029-cursor-narration.md

@@ -0,0 +1 @@
+- fix(cursor): preserve tool context across multi-turn conversations when client lacks conversation_id (#9029)

changelog.d/fixes/9030-antigravity-system-429s.md

@@ -0,0 +1 @@
+- fix(sse): move Antigravity client system content to first user message to avoid upstream 429 RESOURCE_EXHAUSTED on oversized systemInstruction (#9030)

changelog.d/fixes/9073-batches-list-limit-validation.md

@@ -0,0 +1 @@
+- **fix(batches):** `GET /v1/batches` now validates the `limit` query param instead of passing `Number.parseInt(limit)` straight to the SQLite `LIMIT` bind. Previously `?limit=abc` threw an unhandled `datatype mismatch` (→ HTTP 500), `?limit=-1`/`0` returned an incoherent `has_more:true` empty page with `last_id:null`, and a large `?limit` read the entire `batches` table into memory. It now returns a `400` for any non-integer or out-of-range value (1–100, default 20), matching the `POST` handler's Zod validation and the OpenAI Batches contract ([#9073](https://github.com/diegosouzapw/OmniRoute/pull/9073))

changelog.d/fixes/9083-a2a-auth-timing-safe.md

@@ -0,0 +1 @@
+- **fix(a2a):** the A2A JSON-RPC router now compares the bearer token against `OMNIROUTE_API_KEY` in constant time (`crypto.timingSafeEqual`) instead of `===`, closing a token-length timing side-channel, and no longer logs the request URL to server logs on every call ([#9083](https://github.com/diegosouzapw/OmniRoute/pull/9083))

changelog.d/fixes/9088-skills-routes-error-sanitization.md

@@ -0,0 +1 @@
+- **fix(api/skills):** the `/api/skills/**` routes now run caught error messages through `sanitizeErrorMessage` before returning them, so a filesystem failure no longer leaks an absolute path (e.g. `/home/<user>/.omniroute/skills/...`) to the client; the `{ error: string }` response shape is preserved for the dashboard ([#9088](https://github.com/diegosouzapw/OmniRoute/pull/9088))

changelog.d/fixes/9159-fix.plan.md

@@ -0,0 +1 @@
+- fix(management): authorize mcp:connect-only keys on loopback/LAN when requireLogin is enabled (#9159)

changelog.d/fixes/9259-rolling-rpm-leases.md

@@ -0,0 +1 @@
+- **fix(resilience):** Enforce RPM limits with rolling leases and atomic global/provider/account admission ([#9259](https://github.com/diegosouzapw/OmniRoute/pull/9259)). The configured global RPM budget is shared across all enabled provider connections within one process; provider/account overrides add narrower scopes.

changelog.d/fixes/9435-kiro-import-overwrite.md

@@ -0,0 +1 @@
+- fix(oauth): Kiro import token endpoint no longer overwrites existing connection when using shared cached OIDC clientId (#9435)

changelog.d/fixes/9491-port-3005-auth-redirect-login.md

@@ -0,0 +1 @@
+- **fix(auth):** redirect active sessions from /login by checking the session cookie before showing the login form. (thanks @DaDecky)

changelog.d/fixes/9531-ci-combo-matrix.md

@@ -0,0 +1 @@
+- fix(ci): include combo-matrix tests in test-integration job (#9531)

changelog.d/fixes/9534-modelsdevsync-timers.md

@@ -0,0 +1 @@
+- fix(test): prevent flaky modelsDevSync timer assertions by serializing test execution within the file (#9534)

changelog.d/fixes/9536-usage-misreporting.md

@@ -0,0 +1 @@
+- fix(backend): map cache tokens in OpenAI-to-Claude non-streaming usage translation (#9536)

changelog.d/fixes/9541-fastpath-db-corruption.md

@@ -0,0 +1 @@
+- fix(db): add transient-error retry to corruption probe to prevent data loss under concurrent load (#9541)

changelog.d/fixes/9543-searxng-auto-select.md

@@ -0,0 +1 @@
+- fix(search): mark searxng-search as fallbackOnly to prevent auto-select without instance (#9543)

changelog.d/fixes/9545-gpt56-effort-tools.md

@@ -0,0 +1 @@
+- fix(providers): strip provider prefix in getModelTargetFormat to route GPT-5.6 models to /v1/responses (#9545)

changelog.d/fixes/9550-amazon-q-alias.md

@@ -0,0 +1 @@
+- fix(model): add "aq" alias for amazon-q provider so parseModel resolves it instead of falling back to OpenAI (#9550)

changelog.d/fixes/9551-proxyfetch-context-bypass.md

@@ -0,0 +1 @@
+- fix(proxy): NO_PROXY now bypasses context-level proxy in resolveProxyForRequest (#9551)

changelog.d/fixes/9553-prepublish-npm-entry-posix.md

@@ -0,0 +1 @@
+- **fix(build):** `npm run build:cli` (prepublish) no longer fails on POSIX with "npm-cli.js not found next to the running Node binary". The #8858 shim-free npm resolver only knew the Windows layout (`<dir(node.exe)>\node_modules\npm`); on GitHub hosted runners, nvm and system installs npm lives at `<prefix>/lib/node_modules/npm` while node is `<prefix>/bin/node`, so every fresh CI checkout died installing `@omniroute/opencode-plugin` deps (Fast Production Build + dast-smoke red on all PRs). The resolver, extracted to `scripts/build/resolveNpmEntry.ts`, now tries `npm_execpath` (exported by `npm run` itself) first, then the Windows layout, then the POSIX layout — covered by `tests/unit/build/resolve-npm-entry.test.ts` including a live POSIX regression guard.

changelog.d/fixes/9554-file-size-inherited-drift-reconcile.md

@@ -0,0 +1 @@
+- **fix(quality):** reconcile the accumulated file-size drift on `release/v3.8.50` — 13 files sat above their frozen LOC on the clean tip (measured by the gate itself), turning the absolute-mode check (nightly / local) permanently red while the PR-mode base-relative check (#8522) let every innocent PR pass. The per-PR rebaselines were lost across successive conflict resolutions of this hot file during the 08-05/06 merge batch. Frozen values updated to the measured tip for the 11 grown files (each annotated with its owning merged PR: #9024 #9324 #9329 #9193 #9332 #9228 #9260 #8934 #9196 #9163) and `open-sse/executors/default.ts` / `kiro.ts` (above the 1000 cap with no entry) added to the frozen set.

changelog.d/fixes/9559-mcp-audit-vitest-seam.md

@@ -0,0 +1 @@
+- **fix(mcp):** the 3 `audit.test.ts` shutdown/fallback tests fail deterministically since #8959 switched the audit DB loader to `createRequire("better-sqlite3")` — `vi.doMock` only patches Vitest's ESM module graph, so the old better-sqlite3 mock never engaged and the tests hit a real empty sqlite file ("no such table: mcp_tool_audit"), redding the `Vitest (fast-path)` job on every PR (long misdiagnosed as a flake). Shutdown tests now inject the mock through the audit connection cache (`globalThis.__omnirouteMcpAuditDb`), and the node:sqlite fallback test drives a new test-only loader seam (`__setBetterSqliteLoaderForTests`) — the production `createRequire` path is unchanged. 3/3 red → 3/3 green; full `open-sse/mcp-server` vitest suite 88/88.

changelog.d/fixes/9560-turbopack-nft-guard.md

@@ -0,0 +1 @@
+- fix(build): lazy-resolve module-level fs paths to avoid Turbopack NFT whole-source trace (#9560)

changelog.d/fixes/9567-chatcore-sse-flaky.md

@@ -0,0 +1 @@
+- fix(sse): replace timer-based waits with polling to fix flaky chatCore/SSE tests under CI load (#9567)

changelog.d/fixes/9568-gemini-tool-casing.md

@@ -0,0 +1 @@
+- **fix(translator):** restore original tool name casing in Gemini/Antigravity response translators ([#9568](https://github.com/diegosouzapw/OmniRoute/issues/9568))

changelog.d/fixes/9575-tool-call-case.md

@@ -0,0 +1 @@
+- fix(translator): add case-insensitive fallback for upstream tool call name lookups (#9575)

changelog.d/fixes/ghe-copilot-oauth-lifecycle.md

@@ -0,0 +1,2 @@
+- **fix(oauth):** GHE Copilot OAuth lifecycle — connecting an account and refreshing its token both failed. Adding a connection died with `gheUrl is required for GHE Copilot OAuth` because the poll handler's `ghe-copilot` branch was unreachable dead code: the provider is listed in `NO_PKCE_DEVICE_CODE_PROVIDERS`, and that set-based check ran first, calling `pollForToken()` without the `extraData` carrying `gheUrl`. Separately, every manual `Refresh` click surfaced `Token refresh failed — provider returned no new token`, and the proactive pre-request refresh never fired for GHE connections — the manual route, the health-check sweep and `checkAndRefreshToken()` all still special-cased plain `github`, while GHE Copilot's device-code flow never yields a `refresh_token` (only a GitHub access token plus a short-lived Copilot sub-token). `refreshCopilotToken()` now takes an optional `baseUrl` so it can target a GHE host's `<gheUrl>/api/v3` Copilot token endpoint, and `ghe-copilot` is wired in alongside `github` at all four sites.
+- **fix(health-check):** the access-token-only branch of the token health-check sweep no longer logs an unconditional `has no refresh token but has a GitHub access token` line on every tick. That path runs once per 60 s sweep for every `github` / `ghe-copilot` connection, so it emitted ~1440 identical entries per day per connection reporting that nothing had changed. It now logs only when the sweep actually attempted a Copilot sub-token refresh, and says whether that refresh succeeded or failed — so a genuine failure still surfaces instead of being buried in steady-state noise.

changelog.d/fixes/prepublish-native-esbuild.md

@@ -0,0 +1 @@
+- fix(build): exec native esbuild binary directly in prepublish — esbuild ≥0.25 ships an ELF at bin/esbuild and running it through node crashed every build:cli (dast-smoke red on all PRs)

changelog.d/maintenance/9063-cover-specificity-rules.md

@@ -0,0 +1,7 @@
+- **test(sse):** added the first test suite for `open-sse/services/specificityRules.ts` — the
+  module was hotspot #7 in the coverage plan at 11.28% lines with no dedicated test; the 14 pure
+  detectors are now pinned edge-to-edge (token ladders, per-domain max-not-sum scoring, and the
+  min/max clamps), taking the file to ~100% line coverage. The suite also documents two quirks
+  left unchanged: `detectReasoningDepth` scores above 0 on marker-free input via its
+  always-applied message-depth bonus, and `detectErrorContext` returns non-integer scores because
+  it never rounds ([#9063](https://github.com/diegosouzapw/OmniRoute/pull/9063))

changelog.d/maintenance/rtl-physical-class-ratchet.md

@@ -0,0 +1 @@
+- chore(quality): add an RTL layout ratchet — counts physical directional Tailwind classes (ml/mr/pl/pr/left/right/text-left/border-l/rounded-l) that do not mirror under `dir=rtl`, seeded at 1011 so the backlog behind `tests/unit/ui/rtl-logical-classes.test.tsx` ("#3541, partial, core layout") cannot grow while it is worked through

config/quality/complexity-baseline.json

@@ -1,5 +1,8 @@
 {
   "_comment": "Catraca de complexidade (check-complexity.mjs, ESLint core rules complexity>=15 e max-lines-per-function>80 sobre src+open-sse+electron+bin via eslint.complexity.config.mjs). Conta total de violacoes; so pode cair. --update ratcheta.",
+  "_rebaseline_2026_07_25_dario_upstream_proxy_selector": "2130->2175. PR #8523 (Dario embedded service, upstream-proxy mode selector): check:complexity does not run on PR->release fast-gates, so cycle drift accrues unratcheted until a PR trips the gate (same pattern as every _rebaseline_ entry above). Measured base upstream/release/v3.8.49 tip locally at 2169 (with this PR\u0027s own commits removed); this branch measures 2173 local, 2175 on the CI runner (same local-vs-CI off-by-few convention documented in _rebaseline_2026_07_02_v3844_ci_observed). This PR\u0027s own genuine contribution is small (+4 to +6): the new mode <select> branching in ConnectionRow.tsx (Native/CLIProxyAPI/Dario/Fallback + conditional fallback-backend picker) and the probe/adopt/kill-PID branches added to the service supervisor for Dario\u0027s on-demand lifecycle. Using the CI-observed value (2175) so the gate is deterministic where it actually runs, per the established convention. Structural shrink stays tracked in #3501. Tighten via --update next cycle.",
+  "_rebaseline_2026_07_20_owner_night_drain": "Owner-approved (chat, 2026-07-20 ~00:50): 2072->2130. The day's 17 merged PRs consumed the entire slack (tip at 2069/2072); queue PRs #6973(+4)/#7662(+2)/#7719(+1) plus the #7744/#7779 reworks were collectively blocked. Owner chose a wide margin for the remainder of the v3.8.49 cycle instead of per-PR extraction.",
+  "count": 2175,
   "_rebaseline_2026_07_25_8470_hyperagent_sticky_thread": "2130->2170 (+40). PR #8470 (artickc, fix/hyperagent-tool-loop-thread-sticky) pre-green validation. Trust-but-verify: origin/release/v3.8.49 tip alone (pristine, no PR changes) already measures 2169 with node scripts/check/check-complexity.mjs — i.e. +39 is inherited cycle drift unrelated to this PR (the cyclomatic-complexity ratchet does not run on PR->release fast-gates). This PR's OWN growth adds exactly +1: per-file eslint scoped scan on open-sse/executors/hyperagent.ts (base vs PR) shows extractMessageText() crossing the complexity>=15 threshold for the first time (new violation, complexity 25) from the new Anthropic tool_use/tool_result flattening branches; resolveHyperAgentThreadBinding's existing pre-#8470 violation (18) grows to 25 (still counted once, from the new root-key lookup tier); createHyperAgentThread and execute() are unchanged pre-existing violations. Net repo-wide total = 2169 (inherited drift) + 1 (this PR's own new violation) = 2170. Full-repo re-measurement of the merged branch was attempted but not completed live due to heavy concurrent devbox load (many other /green-prs sessions running the identical full-repo eslint scan in parallel); derived from two independently-clean measurements (base-tip full scan + per-file base-vs-PR delta). Covered by tests/unit/executor-hyperagent.test.ts (19/19). Tighten via --update next cycle.",
   "_rebaseline_2026_07_25b_v3849_mergetrain_owngrowth": "Owner-approved (chat, 2026-07-25): 2169->2183 (+14). v3.8.49 /merge-prs 41-PR merge-train aggregate own-growth: measured 2183 on the combined boarded tree (tip ac15014ca7) vs 2169 on the pristine release tip. Each boarded PR sits under the ceiling individually, but the combined batch adds +14 (new branches in #8378 chatCore contextLimit / #8432 cursor native_todo / #8476 combo input-bound / #8526 combo select-all modals / etc \u2014 the pre-screen-flagged complexity-growth set). Same merge-burst-inherited-drift class as the notes below; owner chose absorbing the ceiling over per-PR helper-extraction churn. Structural shrink stays debt (#3501); tighten via --update next cycle.",
   "_rebaseline_2026_07_25_v3849_mergequeue_drain": "Owner-approved (chat, 2026-07-25): 2130->2169 (+39). v3.8.49 /merge-prs queue-drain: the cycle's merge burst (the 8 base-red slices + owner PRs + parallel-session merges #8500-8508) accrued inherited cyclomatic drift the fast-path PR->release never ratchets (check:complexity does not run on PR->release). Measured 2169 on the pristine release tip 4053e2314a alone (BEFORE any queue PR boards) \u2014 so the entire +39 is base drift already on the tip, not any queued PR's own growth. Every merge-ready PR in the queue was tripping Fast Quality Gates on this shared base-red. Owner approved raising the ceiling to the measured tip value so the ~34-PR merge-train lands without per-PR helper-extraction churn. Structural shrink stays debt (#3501); tighten via --update next cycle.",

config/quality/eslint-suppressions.json

@@ -248,11 +248,6 @@
       "count": 1
     }
   },
-  "open-sse/services/rateLimitManager.ts": {
-    "no-restricted-syntax": {
-      "count": 1
-    }
-  },
   "open-sse/services/taskAwareRouter.ts": {
     "@typescript-eslint/no-explicit-any": {
       "count": 4
@@ -903,21 +898,11 @@
       "count": 1
     }
   },
-  "src/app/api/v1/audio/speech/route.ts": {
-    "no-restricted-imports": {
-      "count": 1
-    }
-  },
   "src/app/api/v1/audio/transcriptions/route.ts": {
     "no-restricted-imports": {
       "count": 1
     }
   },
-  "src/app/api/v1/audio/translations/route.ts": {
-    "no-restricted-imports": {
-      "count": 1
-    }
-  },
   "src/app/api/v1/batches/[id]/cancel/route.ts": {
     "no-restricted-imports": {
       "count": 1
@@ -963,11 +948,6 @@
       "count": 2
     }
   },
-  "src/app/api/v1/images/generations/route.ts": {
-    "no-restricted-imports": {
-      "count": 1
-    }
-  },
   "src/app/api/v1/management/proxies/assignments/route.ts": {
     "no-restricted-imports": {
       "count": 1
@@ -2099,11 +2079,6 @@
       "count": 5
     }
   },
-  "tests/unit/combo-cache-invalidation.test.ts": {
-    "@typescript-eslint/no-explicit-any": {
-      "count": 5
-    }
-  },
   "tests/unit/combo-context-length.test.ts": {
     "@typescript-eslint/no-explicit-any": {
       "count": 2
@@ -3229,7 +3204,7 @@
   },
   "tests/unit/translator-claude-to-gemini.test.ts": {
     "@typescript-eslint/no-explicit-any": {
-      "count": 17
+      "count": 15
     }
   },
   "tests/unit/translator-claude-to-openai.test.ts": {
@@ -3347,11 +3322,6 @@
       "count": 2
     }
   },
-  "tests/unit/vertex-functioncall-id-3440.test.ts": {
-    "@typescript-eslint/no-explicit-any": {
-      "count": 2
-    }
-  },
   "tests/unit/vertex-media.test.ts": {
     "@typescript-eslint/no-explicit-any": {
       "count": 3

config/quality/file-size-baseline.json

@@ -1,5 +1,6 @@
 {
-  "_rebaseline_2026_07_24_8470_hyperagent_sticky_thread": "PR #8470 (artickc, fix/hyperagent-tool-loop-thread-sticky) own growth: open-sse/executors/hyperagent.ts 936->1025 (wc -l; check-file-size.mjs counts via split(\"\\n\").length so the gate sees 937->1026, +89, crosses the 1000 cap). Fixes a real bug where a reverse-conversion proxy (text-Intent/JSON to Claude Code native tool_calls) rewrites assistant messages between agentic tool-loop turns, breaking HyperAgent's conversation-prefix fingerprint and cold-starting the thread mid tool-loop. Adds Anthropic tool_use/tool_result flattening to extractMessageText() plus a new rootUserFingerprint()/root-key lookup tier in resolveHyperAgentThreadBinding()/storeHyperAgentThreadAfterTurn() so the thread stays sticky across the tool loop. Cohesive additions inside the existing single-file executor; not extractable without splitting the executor mid-request-flow. Covered by tests/unit/executor-hyperagent.test.ts (19/19, +5 new cases for tool_result/tool_use flattening + root-key stickiness). Pre-merge review flagged a cross-conversation root-key collision risk (tracked in the PR's own mandatory pre-merge checklist, not yet addressed) — unrelated to this file-size ratchet, tracked separately by /fix-prs.",
+  "_rebaseline_2026_08_02_9259_rolling_rpm": "PR #9259 (issue #8733) own growth: open-sse/services/rateLimitManager.ts baseline 1060->1167 (+107; final source 1153). The existing withRateLimit chokepoint now composes process-local rolling RPM leases with Bottleneck admission, releases pre-dispatch leases on queue timeout/abort/connection disable, preserves caller abort reasons, and wires 429/header state into the extracted rollingRpmGate.ts. The remaining growth is irreducible lifecycle wiring at the dispatch boundary plus the real watchdog test hooks needed to verify queued-wedge recovery; moving it further would obscure lease ownership and Bottleneck cleanup. Covered by the focused rate-limit manager/sliding-window suite (33/33); distributed multi-instance coordination remains explicitly out of scope.",
+  "_rebaseline_2026_07_24_8470_hyperagent_sticky_thread": "PR #8470 (artickc, fix/hyperagent-tool-loop-thread-sticky) own growth: open-sse/executors/hyperagent.ts 936->1025 (wc -l; check-file-size.mjs counts via split(\"\\n\").length so the gate sees 937->1026, +89, crosses the 1000 cap). Fixes a real bug where a reverse-conversion proxy (text-Intent/JSON to Claude Code native tool_calls) rewrites assistant messages between agentic tool-loop turns, breaking HyperAgent’s conversation-prefix fingerprint and cold-starting the thread mid tool-loop. Adds Anthropic tool_use/tool_result flattening to extractMessageText() plus a new rootUserFingerprint()/root-key lookup tier in resolveHyperAgentThreadBinding()/storeHyperAgentThreadAfterTurn() so the thread stays sticky across the tool loop. Cohesive additions inside the existing single-file executor; not extractable without splitting the executor mid-request-flow. Covered by tests/unit/executor-hyperagent.test.ts (19/19, +5 new cases for tool_result/tool_use flattening + root-key stickiness). Pre-merge review flagged a cross-conversation root-key collision risk (tracked in the PR’s own mandatory pre-merge checklist, not yet addressed) — unrelated to this file-size ratchet, tracked separately by /fix-prs.",
   "_rebaseline_2026_07_25_8494_capability_filter_fail_closed": "PR #8494 (fix/capability-filters-fail-closed, #8488) own growth: open-sse/services/combo.ts 3640->3693 (+53) adds a fail-closed guard after filterTargetsByRequestCompatibility() — when every eligible target is excluded by request-capability filtering (vision/tools/etc) instead of quota/health, the combo now returns an explicit `capability_mismatch` 400 (describeCapabilityFilterExhaustion, imported from combo/comboStructure.ts) rather than silently falling through to a generic no-targets error, plus a `compatFilterFailOpen` escape hatch (combo config OR settings) mirrored at both the main/auto and round-robin call sites for symmetry. combo/comboStructure.ts (previously under cap, un-frozen) grows 794->918 (+124) — new home for describeCapabilityFilterExhaustion + providerSupportsEmulatedToolCalling (#5240 emulated tool-calling exemption so fail-closed does not regress prompt-emulation-only combos like all-chatgpt-web). Irreducible orchestration wiring at the existing filter chokepoint (same precedent as #7301's universal-cooldown-retry generalization). Companion test tests/unit/combo-routing-engine.test.ts 3409->3449 (+40, fail-closed/fail-open coverage across both call sites) also rebaselined. Covered by tests/unit/8488-capability-filter-fail-closed.test.ts (new) + 95/95 passing across both files. Structural shrink of combo.ts tracked in #3501.",
   "_rebaseline_2026_07_25_8499_ts7_result_union_predicates": "PR #8499 (backryun, chore/ts7-types-executor-scattered) own growth: muse-spark-web.ts 1396->1405 (+9, irreducible). Under this workspace's `strictNullChecks: false`, the boolean-literal discriminant on `GraphqlResult` (`{ ok: true } | { ok: false; error: string }`) narrows the positive `.ok===true` branch but leaves `!result.ok` at the full union under TS7, making `.error` unreachable to the checker at the two call sites (warmup, mode-switch). Fixed by adding a single `isGraphqlFailure()` type-predicate helper (doc comment + 3-line body) reused at both call sites instead of duplicating the predicate inline — not extractable to a shared module without splitting a single-file executor's local narrowing helper out of its own file. Covered by the existing muse-spark-web executor test suite (no behavior change, pure narrowing fix).",
   "_rebaseline_2026_07_22_8131_windowshide_cloudflared_spawn": "PR #8167 (Dingding-leo, fix/windows-hide-child-process, #8131) own growth: src/lib/cloudflaredTunnel.ts 934->935 (+1, irreducible call-site wiring — the single `windowsHide: true` option added to the existing cloudflared spawn() options object so no transient conhost.exe/cmd console window flashes open on Windows). Covered by the pre-merge-fix regression test tests/unit/windows-hide-child-process-spawns-8131.test.ts (added for the two additional spawn() sites the PR missed: ServiceSupervisor.ts, versionManager/processManager.ts) plus the windowsHide assertion added to tests/unit/services/installers/runNpm-shell-5379.test.ts (installers/utils.ts buildNpmExecOptions).",
@@ -13,6 +14,7 @@
   "_rebaseline_2026_07_19_7546_ghe_copilot_route": "PR #7546 (GHE Copilot OAuth provider) own growth: oauth/[provider]/[action]/route.ts 960->963 (gate units, +3 = ghe-copilot device-code wiring at the existing multi-provider device-code branch — reading + HTTPS-validating the gheUrl search param (isValidGheUrl guards at both raw entry points, security-review hardening, 963->970), adding ghe-copilot to the no-PKCE provider set, and building the provider config override / threading gheUrl through poll->postExchange extraData). Mirrors the existing kiro/amazon-q startUrl override pattern right above it in the same branch; cohesive with the existing device-code dispatch chokepoint, not separately extractable without splitting a single provider-switch mid-branch. Frozen so can only shrink; structural shrink tracked in #3501.",
   "_rebaseline_2026_07_19_6846_nvidia_concurrency_gate": "Issue #6846 Phase 1 (nvidia NIM local RPM budget + per-model lockout + per-connection concurrency cap) own growth: open-sse/executors/default.ts 877->890 (+13 = the irreducible call-site wiring at DefaultExecutor.execute(), the only place nvidia requests dispatch through — the existing session-pool body was extracted verbatim into a new private executeWithSessionPool() so the outer execute() can wrap it in the nvidia concurrency-gate acquire/finally-release). All actual gating logic (semaphore key + cap resolution) lives in the new leaf open-sse/executors/default/nvidiaConcurrencyGate.ts (not frozen, well under cap). Covered by tests/unit/nvidia-quota-phase1.test.ts.",
   "_rebaseline_2026_07_18_v3849_provider_detail_wiring": "Merge campaign R2/R3 (2026-07-18): three authorized PRs each add irreducible call-site wiring to ProviderDetailPageClient.tsx — #7360 +5 (ProviderQuotaVisibilityToggle render, component extracted), #7419 +4 (NoAuthProviderControls wiring), #7062 +3 (Dahl provider hook) = 786->798. All three follow the extracted-component pattern (AgentrouterConsoleFields precedent); the frozen file only takes the wiring. Structural shrink tracked in #3501.",
+  "_rebaseline_2026_07_25_dario_upstream_proxy_selector": "PR #8523 (Dario embedded service): upstream-proxy mode selector replaces the binary CLIProxyAPI toggle with Native/CLIProxyAPI/Dario/Fallback + a fallback-backend picker. ProviderDetailPageClient.tsx 798->804 (+6, new hook fields threaded through to ConnectionsListPanel), ConnectionRow.tsx 942->958 (+16, the mode <select> + conditional fallback-backend <select> replacing a single pill button), useProviderConnections.ts 954->986 (+32, upstreamProxyMode/upstreamProxyFallbackBackend state + handleSetUpstreamProxyMode, handleToggleCliproxyapiMode kept as a thin backward-compat wrapper for the existing hook-shape test). All additive UI/state for the new modes — no unrelated refactor.",
   "_rebaseline_2026_07_18_pr7653_chat_tracker_import": "PR #7653 merge-interaction growth: release moved chat.ts to its 1796 cap while this PR adds the single side-effect import 'quotaTrackersBatch.ts' (line 130) — chat.ts IS the canonical quota-fetcher registration point (codex/bailian/deepseek/openrouter/opencode/generic all import+register there), so the +1 is irreducible call-site wiring. 1796->1797. Covered by tests/unit/{agentrouter,v0,freemodel}-quota-fetcher.test.ts.",
   "_rebaseline_2026_07_17_pr7653_agentrouter_console_fields": "PR #7653 own growth (missing acceptance criterion: the AgentRouter quota tracker (#6850) read providerSpecificData.consoleApiKey/newApiUserId but neither field had dashboard UI for provider agentrouter — consoleApiKey was gated to bailian-coding-plan only and newApiUserId had zero UI). AddApiKeyModal.tsx 961->967 (+6) and EditConnectionModal.tsx 1278->1286 (+8) = import + a single <AgentrouterConsoleFields .../> render call plus the newApiUserId formData init field. The actual Input rendering (both consoleApiKey reuse + the new newApiUserId field) was EXTRACTED into a new leaf src/app/(dashboard)/dashboard/providers/[id]/components/modals/AgentrouterConsoleFields.tsx (48 LOC, <cap), mirroring the QuotaScrapingFields.tsx / GlmTeamQuotaFields.tsx precedent (#6351) so the frozen modals only carry the irreducible call-site wiring. Persist logic lives in connectionProviderSpecificData.ts (not frozen). Covered by tests/unit/agentrouter-connection-modal-fields.test.ts.",
   "_rebaseline_2026_07_17_v3849_6842_free_window_wiring": "PR #7651 (openrouter :free-window quota tracking) follow-up: the counter shipped built but never wired into the request pipeline, so combos kept spending guaranteed-429 requests on exhausted free-tier targets. Own growth: src/sse/services/auth.ts 2461->2462 (+1, irreducible at the existing model-aware preflight chokepoint — the `provider === \"codex\"` check that forwards requestedModel into the connection arg is extended to also cover `openrouter`, one added boolean + a doc comment, offset to a single net line by dropping the now-redundant inline condition). Enforcement itself lives in open-sse/services/openrouterQuotaFetcher.ts (not frozen) and the dispatch-time record/correct hooks live in open-sse/executors/base.ts (not frozen). Covered by tests/unit/openrouter-free-window-wiring-6842.test.ts.",
@@ -157,8 +159,188 @@
   "_rebaseline_2026_06_20_1409_1294_models": "Re-baseline src/lib/db/models.ts 1184->1221: combined growth of sibling fixes #1409 (cascade-delete orphaned model aliases when a provider is removed) + #1294 (persist max_input_tokens/max_output_tokens on custom models), both adding CRUD at the existing models domain module. Cohesive db module; not extractable.",
   "_rebaseline_2026_06_20_4389_thinking_toolchoice": "Re-baseline base.ts 1387->1399 (#4389): tool_choice-forced thinking guard at the existing Claude wire-image injection chokepoint (effThinking gate avoids the Anthropic 400 when tool_choice forces a tool). Cohesive guard; structural shrink tracked in #3501.",
   "_rebaseline_2026_07_18_6979_codex_test": "PR #6979 own growth: executor-codex.test.ts 1340->1347 (+7 = generalized ensureThinkingBudget assertion added to the existing codex thinking-budget cases). antigravity-test bump 942->977 REVERTED here: #7408's test split dropped that file to 888, so this PR's +35 fits under the original 942 frozen cap.",
-  "_rebaseline_2026_07_24_8354_logs_timeline_sidebar": "PR #8354 (hartmark, feature/scrolling-log) own growth: src/shared/constants/sidebarVisibility/sections.ts 812->820 (+8, the single new logs-timeline SidebarItemDefinition entry added to LOGS_GROUP.items for the new /dashboard/logs/timeline scrolling request-timeline page). Irreducible data-literal wiring at the existing sidebar-sections chokepoint, same shape as every other item in the file; not extractable without an ad-hoc single-item exception to the file's otherwise-uniform multi-line item style.",
   "cap": 1000,
+  "frozen": {
+    "_rebaseline_2026_07_02_5816_qoder": "PR #5816 (@AgentKiller45, qoder PAT via qodercli): qoderCli.ts 666->989, new-above-cap frozen (owner-approved baseline freeze). The growth is the legitimate PAT job-token exchange + quota parsing CLI transport (the pure-JS Cosy path 500'd on every PAT request); extracting the spawn/parse helpers now would just add indirection to a contributor PR mid-merge. Test frozen also raised for this PR's coverage growth: providers-page-utils.test.ts 1052->1092. Additionally clears an inherited base-red from the already-merged #5933 (codex json_schema->text.format): translator-openai-responses-req.test.ts 1097->1172 (+75 regression tests, no offending branch left). All remain frozen (cannot grow further); release captain's rebaseline-at-release supersedes.",
+    "open-sse/services/qoderCli.ts": 989,
+    "_rebaseline_pr1043_minimax_tts": "Upstream port decolua/9router#1043 (toanalien) own growth: audioSpeech.ts 965->1061 (+96). Adds MiniMax T2A v2 TTS dispatch (handleMinimaxSpeech + hexToBytes helper) — provider entry was already in audioRegistry (format: minimax-tts) but no handler existed, falling through to the OpenAI-compatible default that fails (T2A has custom shape + hex-encoded audio + base_resp envelope). New branch sits next to the other inline provider branches (xiaomi-mimo, coqui, tortoise, aws-polly) — extracting would just create indirection. Covered by tests/unit/minimax-tts-1043.test.ts (3 tests, GREEN: success, base_resp error, invalid-hex).",
+    "_rebaseline_2026_07_23_8266_alibaba_media": "#8266 (@backryun) own growth: imageRegistry.ts 821->979 (+158) — Alibaba-family media models (Qwen image/video, Bailian, Wan) added to the image/video registry. Registry model data, not extractable logic; frozen at new size.",
+    "open-sse/config/imageRegistry.ts": 979,
+    "open-sse/config/providerRegistry.ts": 4731,
+    "open-sse/executors/antigravity.ts": 1813,
+    "open-sse/executors/base.ts": 1540,
+    "open-sse/executors/chatgpt-web.ts": 3206,
+    "open-sse/executors/claude-web.ts": 1057,
+    "open-sse/executors/codex.ts": 1541,
+    "open-sse/executors/cursor.ts": 1577,
+    "open-sse/executors/deepseek-web.ts": 1148,
+    "_rebaseline_2026_06_22_4644_deepseek_web_tools": "PR #4644 (BugsBag/robust deepseek-web tool-call parsing): open-sse/executors/deepseek-web.ts 1117->1125 (+8). The new agentic tool-call path emits surrounding text + reasoning before tool_calls and swaps to the dedicated deepseekWebTools.ts parser; the +8 lines are cohesive wiring at the existing transformSSE chokepoint (the parser itself lives in the new deepseekWebTools.ts file, already under cap). The PR's own fast-gate (PR->release) does not run check:file-size, so this surfaced only at release reconcile. Covered by tests/unit/deepseek-web-tools-variants.test.ts + deepseek-web-tools-execute.test.ts.",
+    "_rebaseline_2026_06_23_4712_deepseek_web_tool_results": "PR for #4712 (deepseek-web drops role:tool): open-sse/executors/deepseek-web.ts 1125->1148 (+23). messagesToPrompt() now folds role:\"tool\" results into the single-prompt transcript (recovering the tool name from the preceding assistant tool_calls by tool_call_id) instead of silently dropping them; the lines are cohesive wiring inside the existing function. Covered by tests/unit/deepseek-web-tool-result-prompt-4712.test.ts.",
+    "_rebaseline_2026_06_28_5237_impersonation_ua_refresh": "PR #5237 (refresh impersonation UAs): grok-web.ts 1871->1873 (+2), muse-spark-web.ts 1284->1302 (+18), perplexity-web.ts 1013->1032 (+19). Net semantic change in each file is a single User-Agent constant (Chrome 147->149 for grok/muse; perplexity kept at Firefox 148 to stay matched with the firefox_148 TLS profile — the contributor's 152 bump was reverted to avoid a UA-vs-JA3 mismatch, #2459). The growth is Prettier reflow that lint-staged unavoidably applies to these grandfathered long-line files the moment they are touched; not extractable. src/sse/services/auth.ts 2336->2401 in the same reconcile is #5222's antigravity-LRU-retry growth that merged via --admin without a baseline bump.",
+    "open-sse/executors/duckduckgo-web.ts": 925,
+    "open-sse/executors/grok-web.ts": 1873,
+    "open-sse/executors/hyperagent.ts": 937,
+    "open-sse/executors/muse-spark-web.ts": 1396,
+    "open-sse/executors/perplexity-web.ts": 1032,
+    "open-sse/handlers/audioSpeech.ts": 1061,
+    "open-sse/handlers/chatCore.ts": 5125,
+    "open-sse/handlers/imageGeneration.ts": 3777,
+    "open-sse/handlers/responseSanitizer.ts": 1139,
+    "open-sse/handlers/search.ts": 1546,
+    "open-sse/handlers/sseParser.ts": 830,
+    "open-sse/handlers/videoGeneration.ts": 1275,
+    "_rebaseline_2026_07_22_8010_codex_responses_engine": "PR #8010 (@JxnLexn) own growth: open-sse/mcp-server/schemas/tools.ts 1497->1505 (+8 = threading the new \"codex-responses\" literal into the compressionConfigureInput strategy/autoTriggerMode Zod enums and setCompressionEngineInput engine enum, mirroring the existing rtk/omniglyph enum entries; no new tool). open-sse/services/compression/strategySelector.ts 1043->1054 (+11 = one new `if (mode === \"codex-responses\")` dispatch branch in runCompression that delegates 100% to the new codexResponsesEngine.apply, mirroring the existing rtk single-mode dispatch, plus threading config.codexResponsesConfig.preserveToolNames into the shared adaptBodyForCompression call at the 3 existing call sites). src/lib/db/compression.ts (untracked, new-file cap 800) 794->845 (+51 = normalizeCodexResponsesConfig, mirroring the existing normalizeRtkConfig normalizer, plus registering \"codex-responses\" in the COMPRESSION_MODES/STACKED_PIPELINE_ENGINE_IDS/SINGLE_MODE_ENGINE sets and the getCompressionSettings load/save switch) — added to the baseline at its current size. All three are cohesive dispatch/normalizer wiring at existing chokepoints (mirroring the prior compression-mode rebaselines #6534/#6556), not extractable without hiding the mode-dispatch boundary. Covered by tests/unit/compression/codex-responses.test.ts (6) + omniglyph-registries.test.ts/types.test.ts (22, updated for the new mode).",
+    "_rebaseline_2026_07_22_8034_compression_exclusions_persistence": "#8034 (compression exclusions) own growth: src/lib/db/compression.ts 845->850 (+5 = threading the new compressionExclusions field through the existing getCompressionSettings/saveCompressionSettings load/save switch over the shared key_value compression namespace — no new table, no raw SQL). Mirrors the prior compression-field rebaselines (#8010 codex-responses normalizer at the same chokepoint); the load/save switch is a single dispatch boundary, not extractable without hiding it. Covered by the PR's 8 node:test + 3 vitest cases.",
+    "src/lib/db/compression.ts": 866,
+    "open-sse/mcp-server/schemas/tools.ts": 1505,
+    "open-sse/mcp-server/server.ts": 1555,
+    "open-sse/mcp-server/tools/advancedTools.ts": 1120,
+    "_rebaseline_2026_06_27_5193_antigravity_basered": "Base-red (pre-existing release drift, fast-gate PR->release skips check:file-size): accountFallback.ts 1773->1777 and src/app/api/providers/[id]/test/route.ts 924->940 were already over their frozen caps on release/v3.8.39 independent of any antigravity change. Owner chose to rebaseline (keep the documented issue-reference comments #1846/#1449/#347 etc.) rather than accept the contributor comment-stripping in #5200/#5198. Reverted #5200 to restore the comments; bumped these two frozen caps to the actual base sizes. No logic change.",
+    "_rebaseline_2026_07_22_8213_gemini_tpm_quota_cooldown_wait": "PR #8213 (hartmark, fix/gemini-tpm-quota-cooldown-wait) own growth: open-sse/services/accountFallback.ts 1857->1932 on the merged tip (release 1892 incl #8050 +35, plus this PR own growth +40); measured against the PR own merge-base was 1857->1898 (+41 — the release tip separately carries an unrelated +34 from #8050's antigravity 404 model-not-found lockout scoping, which this PR's branch does not include and this entry does not cover). Own growth is the Gemini TPM-ceiling classification + cooldown-wait wiring feeding into the combo cooldown-wait state machine (rate-limit wedge recovery) introduced by this PR's commit series. Irreducible additions at the existing account-fallback/model-lockout chokepoint. Covered by the PR's own gemini-rate-limit-tracker and TPM-ceiling benchmark test additions.",
+    "_rebaseline_2026_07_23_8252_combo_400_advance": "#8252 (@RaviTharuma) own growth: accountFallback.ts 1932->1940 (+8) + combo.ts 3604->3630 (+26) — advance combo on model-scoped 400s wrapped as invalid/Bad-Request. Irreducible wiring at existing account-fallback + combo dispatch chokepoints. Covered by combo-model-scoped-400-advance.test.ts.",
+    "open-sse/services/accountFallback.ts": 1940,
+    "open-sse/services/adobeFireflyClient.ts": 1958,
+    "open-sse/services/batchProcessor.ts": 915,
+    "open-sse/services/browserBackedChat.ts": 850,
+    "open-sse/services/claudeCodeCompatible.ts": 1202,
+    "_rebaseline_pr4592_exclude_exhausted_auto": "Reconcile #4592 already-merged growth: combo.ts 2991->3036 (+45, terminal-status quota-cutoff exclusion in buildAutoCandidates + opt-in gate). Fast-gate PR->release does not run check:file-size.",
+    "_rebaseline_2026_06_24_headroom_strategy": "Headroom-aware connection selection (dario technique): combo.ts 3168->3180 (+12 = a new `else if (strategy === \"headroom\")` dispatch branch in handleComboChat that delegates to orderTargetsByHeadroom + its log line, plus the import). The actual logic lives OUT of the god-file: the pure ranker rankByHeadroom/computeHeadroom is the new leaf open-sse/services/combo/headroomRanking.ts (91 LOC, <cap) and the async orderer orderTargetsByHeadroom is appended to the existing open-sse/services/combo/quotaStrategies.ts (<cap) next to its sibling reset-aware/reset-window orderers (reuses their connection-expansion machinery). headroom = 1 - max(util_5h, util_7d) from getSaturation (src/lib/quota/saturationSignals.ts), prefers the connection with the most free capacity. Only the dispatch wiring is irreducible at the existing combo strategy chokepoint (mirrors the reset-aware/reset-window/context-optimized branches); not extractable without hiding the call site. fill-first stays default; all existing strategies untouched. Covered by tests/unit/combo-headroom-ranking.test.ts (pure helper) + tests/unit/combo-headroom-strategy.test.ts (orderer, saturation injected). Structural shrink of combo.ts tracked in #3501.",
+    "_rebaseline_2026_06_24_quota_share_strategy": "Dedicated quota-share strategy (Phase 3 #9): combo.ts 3180->3190 (+10 = one new `else if (strategy === \"quota-share\")` dispatch branch in handleComboChat that delegates 100% to selectQuotaShareTarget + its log line, plus the import). All the new logic lives OUT of the god-file in two new leaves under open-sse/services/combo/: quotaShareInflight.ts (in-flight counter with TTL/lease, ~150 LOC <cap) and quotaShareStrategy.ts (per-model bucket gating via isBucketSaturated + DRR proportional to weight + P2C over in-flight, ~240 LOC <cap). Only the dispatch wiring is irreducible at the existing combo strategy chokepoint (mirrors the headroom/reset-aware/reset-window/context-optimized branches); not extractable without hiding the call site. ZERO existing strategy cases were modified — only this branch was added, and the qtSd/ combos switched from fill-first to quota-share in src/lib/quota/quotaCombos.ts. Covered by tests/unit/quota-share-strategy.test.ts (gating, DRR fairness, P2C in-flight, fail-open, activation). Structural shrink of combo.ts tracked in #3501.",
+    "_rebaseline_2026_06_24_task_aware_routing": "Task-aware routing strategy (port PR #2045, OmniRoute #4945): combo.ts 3190->3225 (+35) = one new `else if (strategy === \"task-aware\")` dispatch branch delegating 100% to selectTaskAwareTarget + its imports/log lines. All scoring/classification logic lives OUT of the god-file in the new leaf open-sse/services/taskAwareRouting.ts (553 LOC <cap). Only the dispatch wiring is irreducible at the existing combo strategy chokepoint (mirrors quota-share/headroom/reset-aware branches). ZERO existing strategy cases modified. Covered by tests/unit/combo-task-aware.test.ts (35 tests). Structural shrink of combo.ts tracked in #3501.",
+    "_rebaseline_2026_07_22_8213_combo_cooldown_wait_recording": "PR #8213 (hartmark, fix/gemini-tpm-quota-cooldown-wait) own growth: open-sse/services/combo.ts 3548->3604 (+56, entirely this PR's diff — no other commit touched this file between the PR's merge-base and the release tip). Fixes combo cooldown-wait state recording so a bogus 503 is no longer crystallized when the cooldown-wait vars reset every setTry, adds an OpenAI-format SSE error frame path for combo-exhausted rejections (capturing request body + attempted models), and gives an abandoned per-target dispatch its own timeout instead of leaking a permanent 'pending' dashboard entry. Irreducible additions at the existing handleComboChat dispatch/retry chokepoint (mirrors the prior quota-share/headroom/task-aware strategy-branch precedents already frozen in this file). Covered by the PR's own combo-config + Gemini TPM-ceiling benchmark test additions.",
+    "open-sse/services/combo.ts": 3630,
+    "_rebaseline_2026_06_26_fidelity_gate_extraction": "Milestone-B fidelity-gate wiring residual: bodyToText+gateAdvance extracted to fidelityGateStep.ts (889->854, -35), but the StackOptions.fidelityGate field, the `const fidelityGate` reads at the two stacked-loop dispatch chokepoints, and the import of FidelityGateConfig are irreducible wiring that cannot leave strategySelector without an architectural refactor of the pre-existing stacked pipeline. Net: 889->854 (+6 vs the pre-Milestone-B frozen 848). Covered by tests/unit/compression/*.test.ts (940 pass).",
+    "_rebaseline_2026_06_28_5243_risk_gate_prepass": "PR #5243 (compression risk-gate pre-pass) own growth: open-sse/services/compression/strategySelector.ts 854->899 (+45). The three exported entry points (applyCompression/applyStackedCompression/applyStackedCompressionAsync) become thin wrappers over pure-extracted private bodies (runCompression/runStackedCompression/runStackedCompressionAsync) so the risk-gate mask->run->restore wrapper sits strictly OUTSIDE the per-step loop — a single universal integration point. The wrapper logic itself (resolveRiskGate/withRiskGate) lives in the new riskGate/strategyWrap.ts (<cap); the residual growth is the duplicated thin-wrapper signatures + the extracted bodies' dispatch boundary, guarded by a byte-identical parity test (riskGateIntegration). Default off (DEFAULT_COMPRESSION_CONFIG unchanged). Not extractable without hiding the dispatch boundary, mirroring prior compression rebaselines. Structural shrink tracked in #3501.",
+    "_rebaseline_2026_06_29_5286_memoization": "PR #5286 own growth: strategySelector.ts 899->960 (+61 = the opt-in result-memoization branches in applyCompression/applyCompressionAsync — principal+determinism gate, makeMemoKey lookup/store with model+supportsVision folded into the key, recompute-with-memo-off). Default off (memoizeCompressionResults), so zero behavior change. The memo helpers live in the leaf resultMemo.ts (<cap); the chokepoint wiring here is not extractable. Structural shrink of this hot-path file tracked in #3501.",
+    "open-sse/services/compression/strategySelector.ts": 1060,
+    "_rebaseline_2026_07_10_gcf_v3_2_decode": "PR #6838 own growth: new vendored file open-sse/services/compression/engines/headroom/gcf/decode_generic.ts frozen at 880 (> 800 cap). It is the vendored GCF generic-profile decoder (spec v3.2 nested flattening plus the prototype-pollution / hasOwnProperty hardening added in this PR's Gemini review). Kept as one file faithful to upstream gcf-typescript so re-vendoring stays a clean copy rather than a re-split each cycle (sibling generic.ts/scalar.ts stay < cap; extraction would also fragment the file's frozen eslint no-explicit-any suppressions). Round-trip + prototype-pollution regression coverage in tests/unit/compression/headroom-smartcrusher.test.ts. Frozen: only shrinks from here.",
+    "open-sse/services/compression/engines/headroom/gcf/decode_generic.ts": 880,
+    "open-sse/services/rateLimitManager.ts": 1035,
+    "_rebaseline_2026_06_29_4038_cas_guard": "PR (#4038) own growth: tokenRefresh.ts 2103->2181 (+78 = the compare-and-swap guard on the refresh persist — runWithCasGuard/getActiveCasGuard AsyncLocalStorage pair mirroring runWithOnPersist, casGuardShouldSkipPersist that rereads the row right before persisting and skips the write when a concurrent writer already rotated the refresh_token past the one presented, plus getCasGuardStats counters). Fixes the sibling-rotation-revert → token-family-revocation storm. Gated behind an active guard (opt-in; no guard => byte-identical). Wiring lives at the two persist chokepoints inside getAccessToken; the comparison reuses wasRefreshTokenRotated from refreshSerializer. Not extractable without splitting the refresh hot path.",
+    "_rebaseline_2026_07_09_6126_clinepass_dual_auth": "PR #6126 (@hajilok, dual-auth ClinePass) own growth: tokenRefresh.ts 2181->2182 (+1 = a single `case \"clinepass\":` fallthrough label added to the existing `case \"cline\":` in _getAccessTokenInternal's provider switch, so clinepass token refresh dispatches to the already-shared refreshClineToken() instead of silently falling through to the generic OAuth refresh). Irreducible 1-line switch-case wiring at the existing chokepoint; the header-building logic for the same feature was extracted to a new leaf src/shared/utils/clineAuth.ts::buildClinepassHeaders() (well under cap) to avoid growing open-sse/executors/default.ts. Covered by tests/unit/clinepass-provider.test.ts.",
+    "_rebaseline_2026_07_09_6363_kiro_external_idp": "PR #6363 (@artickc, Kiro external IdP) own growth: tokenRefresh.ts 2182->2249 (+67 = the external_idp refresh branch inside refreshKiroToken — standard public-client OAuth2 refresh_token grant against the org IdP tokenEndpoint via buildExternalIdpRefreshParams/isExternalIdpAuthMethod from the new leaf open-sse/services/kiroExternalIdp.ts, with invalid_grant/invalid_client -> unrecoverable_refresh_error mapping). Cohesive addition at the existing refreshKiroToken chokepoint. Covered by tests/unit/kiro-external-idp.test.ts.",
+    "open-sse/services/tokenRefresh.ts": 2249,
+    "open-sse/services/usage.ts": 3454,
+    "open-sse/translator/request/openai-to-gemini.ts": 906,
+    "open-sse/translator/request/openai-to-kiro.ts": 912,
+    "_rebaseline_2026_07_22_8211_gemini_malformed_tool_choice": "PR #8211 (hartmark, fix/gemini-malformed-function-call-tool-choice) own growth: open-sse/translator/response/gemini-to-openai.ts 771->821 (+50, entirely this PR's diff — no other commit touched this file between the PR's merge-base and the release tip). Adds MALFORMED_FUNCTION_CALL/UNEXPECTED_TOOL_CALL handling inside geminiToOpenAIResponse(): synthesizes a `malformed_tool_call` tool_calls entry so finish_reason normalizes to the standard \"tool_calls\" instead of an unrecognized raw enum value that OpenAI-compatible clients (e.g. OpenClaw) silently ignore, and always synthesizes (rather than skipping when a real tool call already exists) so a malformed attempt alongside a real one in the same turn is not silently discarded. Irreducible cohesive addition at the existing candidate/finishReason translation chokepoint (mirrors the 9router#2462 raw-finish-reason precedent immediately below it in the same function). Covered by the PR's own tests/unit test additions for both the malformed-only and malformed-plus-real-call cases.",
+    "open-sse/translator/response/gemini-to-openai.ts": 821,
+    "_rebaseline_2026_07_22_7936_namespace_roundtrip": "#7936 (@RCrushMe, Responses-Chat namespace round-trip identity seam) own growth: open-sse/translator/response/openai-responses.ts 1092->1125 (+33) and open-sse/utils/stream.ts 2814->2869 (+55) — threading the namespace-identity seam through the Responses↔Chat translation + stream paths so tool-call namespaces survive the round-trip. Cohesive translation/stream wiring at existing chokepoints, frozen at new size.",
+    "_rebaseline_2026_07_22_8210_openrouter_midstream_error": "PR #8210 (hartmark, fix/openrouter-midstream-error-surfacing) own growth: open-sse/translator/response/openai-responses.ts 1137->1163 (+26) measured on the merged tip (release 1137 + this PR own growth). Adds a single new branch inside openaiToOpenAIResponsesResponse() that detects an OpenRouter-style mid-stream aggregator error (HTTP 200 SSE chunk with empty choices + a top-level error object) and surfaces it as state.upstreamError instead of silently falling through to the no-op/awaitingTrailingUsage path, which previously masked the failure as a false empty-success completion and skipped combo fallback. Irreducible call-site addition at the existing chunk-dispatch chokepoint (mirrors the Gemini-to-OpenAI translator's #4177 precedent for the same class of upstream error surfacing). Note: this baseline entry does NOT cover the separate pre-existing +11 drift already on the release tip from #8081/#8162 (1125->1136, unrelated reasoning-placeholder-stripping fix merged after this PR branched) — that drift belongs to the maintainer's rebaseline, not this PR.",
+    "open-sse/translator/response/openai-responses.ts": 1163,
+    "open-sse/utils/cursorAgentProtobuf.ts": 1521,
+    "_rebaseline_2026_07_23_8143_empty_catch_logging": "#8143 (@chirag127) own growth: open-sse/utils/stream.ts 2869->2887 (+18) — replacing empty catch blocks in the SSE stream subsystem with console.debug logging (Rule #6 silent-swallow fix, issues #8138-#8142). Cohesive logging additions at the existing catch chokepoints, not extractable; frozen at new size. Covered by tests/unit/stream-handler-catch-logging-8143.test.ts.",
+    "open-sse/utils/stream.ts": 2887,
+    "src/app/(dashboard)/dashboard/HomePageClient.tsx": 1385,
+    "src/app/(dashboard)/dashboard/analytics/ComboHealthTab.tsx": 1031,
+    "src/app/(dashboard)/dashboard/api-manager/ApiManagerPageClient.tsx": 3120,
+    "src/app/(dashboard)/dashboard/cache/media/MediaPageClient.tsx": 1105,
+    "src/app/(dashboard)/dashboard/cache/page.tsx": 845,
+    "src/app/(dashboard)/dashboard/cli-code/components/CodexToolCard.tsx": 900,
+    "src/app/(dashboard)/dashboard/cloud-agents/page.tsx": 931,
+    "_rebaseline_2026_07_15_7070_combos_memo": "PR #7070 (perf/p1-memo) own growth: src/app/(dashboard)/dashboard/combos/page.tsx 4655->4656 (+1 = React.memo wrapping of ComboCard). Covered by tests/unit/ui/combos-page-smoke.test.tsx.",
+    "src/app/(dashboard)/dashboard/combos/page.tsx": 4656,
+    "src/app/(dashboard)/dashboard/costs/CostOverviewTab.tsx": 1495,
+    "src/app/(dashboard)/dashboard/costs/quota-share/components/PoolWizard.tsx": 1022,
+    "src/app/(dashboard)/dashboard/endpoint/EndpointPageClient.tsx": 2615,
+    "_rebaseline_2026_07_22_8213_health_unblock_model_cooldowns": "PR #8213 (hartmark, fix/gemini-tpm-quota-cooldown-wait) own growth: src/app/(dashboard)/dashboard/health/page.tsx 1094->1165 (+71, entirely this PR's diff — no other commit touched this file between the PR's merge-base and the release tip). Adds handleUnblockAll/handleUnblockOne dashboard actions (DELETE /api/resilience/model-cooldowns) so an operator can manually clear a Gemini TPM-wedge model lockout surfaced by this PR's cooldown-wait fixes, instead of waiting out the ceiling. Irreducible UI wiring at the existing health-page action chokepoint. Covered by the PR's own dashboard/resilience test additions.",
+    "src/app/(dashboard)/dashboard/health/page.tsx": 1165,
+    "src/app/(dashboard)/dashboard/playground/components/tabs/ApiTab.tsx": 847,
+    "src/app/(dashboard)/dashboard/providers/[id]/ProviderDetailPageClient.tsx": 804,
+    "src/app/(dashboard)/dashboard/providers/[id]/components/ConnectionRow.tsx": 958,
+    "src/app/(dashboard)/dashboard/providers/[id]/components/modals/AddApiKeyModal.tsx": 967,
+    "src/app/(dashboard)/dashboard/providers/[id]/components/modals/EditConnectionModal.tsx": 1288,
+    "src/app/(dashboard)/dashboard/providers/[id]/hooks/useProviderConnections.ts": 986,
+    "src/app/(dashboard)/dashboard/providers/[id]/hooks/useProviderModels.ts": 155,
+    "src/app/(dashboard)/dashboard/providers/[id]/hooks/useProviderSettings.ts": 264,
+    "src/app/(dashboard)/dashboard/providers/[id]/providerPageHelpers.ts": 1054,
+    "src/app/(dashboard)/dashboard/providers/components/onboarding/ProviderOnboardingWizard.tsx": 948,
+    "src/app/(dashboard)/dashboard/providers/page.tsx": 1927,
+    "src/app/(dashboard)/dashboard/runtime/RuntimePageClient.tsx": 1201,
+    "src/app/(dashboard)/dashboard/settings/components/AppearanceTab.tsx": 819,
+    "src/app/(dashboard)/dashboard/settings/components/ComboDefaultsTab.tsx": 903,
+    "src/app/(dashboard)/dashboard/settings/components/CompressionSettingsTab.tsx": 974,
+    "src/app/(dashboard)/dashboard/settings/components/MemorySkillsTab.tsx": 898,
+    "src/app/(dashboard)/dashboard/settings/components/PricingTab.tsx": 1019,
+    "src/app/(dashboard)/dashboard/settings/components/ProxyRegistryManager.tsx": 1464,
+    "src/app/(dashboard)/dashboard/settings/components/ResilienceTab.tsx": 1183,
+    "src/app/(dashboard)/dashboard/settings/components/RoutingTab.tsx": 1629,
+    "src/app/(dashboard)/dashboard/settings/components/SystemStorageTab.tsx": 1924,
+    "src/app/(dashboard)/dashboard/usage/components/BudgetTab.tsx": 1028,
+    "src/app/(dashboard)/dashboard/usage/components/EvalsTab.tsx": 2148,
+    "src/app/(dashboard)/dashboard/usage/components/ProviderLimits/index.tsx": 1127,
+    "src/app/api/oauth/[provider]/[action]/route.ts": 970,
+    "src/app/api/providers/[id]/models/route.ts": 2593,
+    "src/app/api/providers/[id]/test/route.ts": 940,
+    "src/app/api/usage/analytics/route.ts": 948,
+    "src/app/api/v1/models/catalog.ts": 1615,
+    "src/lib/cloudflaredTunnel.ts": 935,
+    "src/lib/db/apiKeys.ts": 1662,
+    "src/lib/db/core.ts": 1825,
+    "src/lib/db/migrationRunner.ts": 1125,
+    "src/lib/db/models.ts": 1259,
+    "src/lib/db/providers.ts": 1107,
+    "src/lib/db/proxies.ts": 1177,
+    "src/lib/db/settings.ts": 1155,
+    "src/lib/db/usageAnalytics.ts": 925,
+    "src/lib/evals/evalRunner.ts": 961,
+    "src/lib/memory/retrieval.ts": 1171,
+    "src/lib/modelsDevSync.ts": 934,
+    "src/lib/providers/validation.ts": 4523,
+    "src/lib/resilience/settings.ts": 841,
+    "src/lib/tailscaleTunnel.ts": 1202,
+    "src/lib/usage/callLogs.ts": 997,
+    "src/lib/usage/providerLimits.ts": 1006,
+    "src/lib/usage/usageHistory.ts": 988,
+    "_rebaseline_2026_06_27_5193_5203_antigravity_oauthmodal": "Antigravity remote-login own growth: OAuthModal.tsx 960->969 (gate units). #5193 (+~4: remote paste instruction shown for all remote incl. Google + its rationale comment) and #5203 (+~5: handleManualSubmit credential-blob branch + button guard; submit logic extracted to oauthBlobSubmit.ts to minimize). Frozen set to the SUM so either merge order passes. Cohesive at the existing manual-submit chokepoint.",
+    "_rebaseline_2026_07_18_7399_xai_oauth_modal": "PR #7399 (xAI OAuth PKCE) own growth: OAuthModal.tsx 993->998 (+5 = provider entry + PKCE flow branch wiring at the existing provider-switch chokepoint; the provider logic itself lives in src/lib/oauth/providers/xai-oauth.ts, new leaf). Third irreducible wiring bump on this modal (969->989->993->998); structural shrink tracked in #3501.",
+    "_rebaseline_2026_07_19_6636_codex_session_json": "#6636 own growth: OAuthModal.tsx 998->1030 (gate units, split(\"\\n\").length incl. trailing newline; +32 = session-JSON paste branch for handleManualSubmit plus a shared submitCodexAccessToken() helper extracted from the pre-existing bare-JWT branch, mirroring the #5203 oauthBlobSubmit.ts extraction precedent; the normalizer logic itself lives in the new src/lib/oauth/utils/codexSessionImport.ts leaf module, not here). Fourth irreducible wiring bump on this modal (969->989->993->998->1030); structural shrink tracked in #3501.",
+    "_rebaseline_2026_07_19_7546_ghe_copilot_modal": "PR #7546 (GHE Copilot OAuth provider) own growth: OAuthModal.tsx 1030->1056 (gate units). Adds a gheUrl input state, routes ghe-copilot through the existing device-code branch, and threads gheUrl into the device-code request/poll extraData at the existing provider-switch chokepoints (+~24 lines, cohesive with the same pattern as #7399/#6636). The standalone GHE enterprise-URL config step JSX (originally +31 lines inline) was extracted to the new src/shared/components/oauthModal/GheConfigStep.tsx leaf component to minimize the bump; what remains is the irreducible provider-branch wiring. Fifth bump on this modal (969->989->993->998->1030->1056); structural shrink tracked in #3501.",
+    "_rebaseline_2026_07_21_8027_grok_cli_auth_json_paste": "PR #8027 (RaviTharuma, fix(grok-cli) #7610) own growth: OAuthModal.tsx 1080->1100 (gate units). Requires the full ~/.grok/auth.json (with refresh_token) on the paste-import path instead of a bare JWT, at the existing paste-token chokepoint (renamed tab label, updated instructions/placeholder, textarea for the auth.json blob, inline error surface). The validation logic itself (parseGrokCliPasteToken, previously an inline ~75-line function) was extracted to the new src/lib/oauth/utils/grokCliAuthJson.ts leaf module — mirroring the #6636/#7546 extraction precedent — so only the irreducible UI wiring remains here. Sixth bump on this modal (969->989->993->998->1030->1056->1100); structural shrink tracked in #3501.",
+    "src/shared/components/OAuthModal.tsx": 1100,
+    "_rebaseline_2026_07_22_8213_requestloggerdetail_unblock_ui": "PR #8213 (hartmark, fix/gemini-tpm-quota-cooldown-wait) own growth: src/shared/components/RequestLoggerDetail.tsx 799->941 (+142, entirely this PR's diff — no other commit touched this file between the PR's merge-base and the release tip; crosses the general 800-line new-file cap so is frozen here for the first time). Adds a collapsible section header (open/expand-less toggle) plus per-log-entry unblock (`unblocking`/`cleared` state, isCombo503 detection) so the request-logger detail panel surfaces the same Gemini TPM cooldown-wait / model-lockout unblock action introduced by this PR at the individual-request level (mirrors the health-page bulk unblock action added in the same PR). Covered by the PR's own dashboard/resilience test additions.",
+    "src/shared/components/RequestLoggerDetail.tsx": 941,
+    "src/shared/components/RequestLoggerV2.tsx": 1629,
+    "src/shared/components/analytics/charts.tsx": 1558,
+    "_rebaseline_2026_07_10_6318_omp_letta": "PR #6318 (@hamsa0x7, omp+letta CLI integrations) own growth: cliTools.ts (+53 = 2 registry entries incl. omp docsUrl) and cliRuntime.ts (+18 = runtime-detection wiring for the 2 new tools). Cohesive registry/wiring growth at the existing chokepoints; scope reduced from the original 5 tools (pi/codewhale/jcode shipped separately).",
+    "src/shared/constants/cliTools.ts": 916,
+    "src/shared/constants/pricing.ts": 1662,
+    "src/shared/constants/providers.ts": 3276,
+    "src/shared/constants/sidebarVisibility.ts": 1198,
+    "src/shared/services/cliRuntime.ts": 1128,
+    "src/shared/validation/schemas.ts": 2523,
+    "_rebaseline_2026_06_28_5275_correlation_id_extract": "Extraction of the safe CorrelationId subset of #5275 (hartmark) — request correlation id stored in call_logs (migration 109) and returned via the X-Correlation-Id response header, WITHOUT the combo/resilience or build/lazy-loading changes (those stay in #5275). Own growth: callLogs.ts 975->985 (correlation_id column on CallLogSummaryRow + read/map), usageHistory.ts 983->988 (correlationId metadata normalize), chat.ts 1575->1632 (withCorrelationId response wiring + combo-failure log carrying correlationId), chatHelpers.ts new 811 (withCorrelationId helper + reqId threading; was 791<cap pre-feature). Cohesive request/logging chokepoint wiring; structural shrink of chat.ts tracked in #3501.",
+    "_rebaseline_2026_07_09_6678_routing_strategy_9router": "#6678 (SeaXen) — 9router-parity Routing Strategy settings card + per-provider/combo sticky-round-robin override. Own growth: ProviderDetailPageClient.tsx 784->786 (single ProviderAccountRoutingCard mount + import), auth.ts 2448->2458 (providerStrategies override resolution: fallbackStrategy/stickyRoundRobinLimit per-provider cascade in getProviderCredentials). Both additive, zero unrelated refactor; new UI/logic lives in new files (ProviderAccountRoutingCard.tsx, RoutingStrategyCard.tsx, rrState.ts::resolveComboStickyRoundRobinLimit). chat.ts value below reflects the current release tip (grown by other concurrent PRs, e.g. #6640), not this PR own change.",
+    "_rebaseline_2026_07_22_8213_chat_abandoned_target_abort": "PR #8213 (hartmark, fix/gemini-tpm-quota-cooldown-wait) own growth: src/sse/handlers/chat.ts 1794->1860 (+66, measured against the PR's own merge-base — the release tip separately carries an unrelated -5 net shrink from #8013's antigravity callable-catalog alignment, which this PR's branch does not include and this entry does not cover). Adds resolveDispatchClientRawRequest(): merges a per-target modelAbortSignal into clientRawRequest.signal (via mergeAbortSignals) so a combo target abandoned by comboTargetTimeoutMs actually observes its own abort and reaches its cleanup path, instead of hanging forever inside withRateLimit/acquireAccountSemaphore and leaking a permanent 'pending' dashboard entry (live incident, log id 1784418258231-14961a). Also wires combo-exhausted rejection logging to capture request body + attempted models via the new rejectedRequestUsage helper. Irreducible additions at the existing chat dispatch chokepoint. Covered by the PR's own combo-config + integration test additions.",
+    "_rebaseline_2026_07_23_8127_grok_weekly_quota": "#8127 (@apoapostolov) own growth: src/sse/handlers/chat.ts 1861->1865 (+4) — weekly quota tracking for grok-web wires a quota-fetch hook at the existing dispatch chokepoint. Thin wiring mirroring adjacent provider-quota branches; not extractable. Covered by tests/unit/grok-quota-fetcher.test.ts.",
+    "src/sse/handlers/chat.ts": 1865,
+    "_rebaseline_2026_07_20_7779_routingcombo_thread": "PR #7779 own growth: chatHelpers.ts 876->877 (+1, thread routingComboId into executeChatWithBreaker for compression-combo assignment). Frozen so it can only shrink.",
+    "src/sse/handlers/chatHelpers.ts": 878,
+    "src/sse/services/auth.ts": 2475,
+    "open-sse/executors/default.ts": 890,
+    "open-sse/translator/request/openai-responses.ts": 902,
+    "open-sse/executors/kiro.ts": 944,
+    "open-sse/translator/request/openai-to-claude.ts": 823,
+    "tests/unit/account-fallback-service.test.ts": 1572,
+    "tests/unit/provider-validation-specialty.test.ts": 2980,
+    "open-sse/executors/huggingchat.ts": 813,
+    "_rebaseline_2026_07_01_v3843_release_5609": "Rebaseline v3.8.43 (PR #5609 release reconciliation). DRIFT dos 109 commits do ciclo: 8 god-files existentes cresceram (ApiManagerPageClient 2983->3017, combos/page 4594->4608, AddApiKeyModal 868->869, providerPageHelpers 974->996, chat.ts 1635->1647, auth.ts 2401->2403, batchProcessor 828->915, combo.ts 3368->3387) + 2 novos acima do cap (huggingchat.ts 813, tests web-cookie-providers-new 827) + 4 test files cresceram. Modularizacao deferida (blast-radius mid-release); congelado no estado atual p/ o proximo ciclo ratchetar daqui.",
+    "src/lib/providers/validation/webProvidersA.ts": 809,
+    "src/lib/tokenHealthCheck.ts": 832,
+    "_rebaseline_2026_07_09_6587_kiro_api_key_auth": "PR #6587 (@strangersp) own growth for Kiro long-lived API-key auth, merged onto v3.8.47 tip: openai-to-kiro.ts 890->912 (+22, auth-header selection for API-key-vs-OAuth-token connections), providerLimits.ts 998->1000 (+2, API-key auth-type branch), translator-openai-to-kiro.test.ts 1234->1257 (+23), providers-page-utils.test.ts 1109->1107 (net -2 after merging with parallel release drift; connectionMatchesProviderCard api_key coverage added), provider-validation-specialty.test.ts 2856->2980 (+124 net after merge with parallel release drift; this PR also removed the file's `@typescript-eslint/no-explicit-any` eslint-suppression entry by fixing all `any` usages, adding typed replacements). Cohesive additive feature growth, well tested; not extractable without splitting the existing chokepoints mid-merge.",
+    "_rebaseline_2026_07_19_7787_ic2_localdb_reexports": "PR #7787 (IC2 raw connections cache + lazy-decrypt) own growth: localDb.ts 805->807 (gate units, +2). localDb.ts is the re-export-only layer (hard rule #2 — no logic); the PR adds 4 new db/readCache re-exports (touchConnectionLastUsed, getCachedRawProviderConnections, getCachedProviderConnectionById, getCachedProviderNodes) required by existing barrel importers. Irreducible for a re-export list; frozen so it can only shrink.",
+    "_rebaseline_2026_07_20_7819_autocandidateoverrides_reexport": "PR for #7819 (Level 1+2: read-only auto/* candidate transparency + per-API-key exclusions) own growth: localDb.ts 807->808 (+1). Adds a single `export * from \"./db/autoCandidateOverrides\"` barrel re-export (hard rule #2 — no logic) for the new DB module backing per-apiKey candidate exclusions. Irreducible for a re-export list; frozen so it can only shrink.",
+    "src/lib/localDb.ts": 808,
+    "_rebaseline_2026_07_12_v3847_mergeprs_tail": "v3.8.47 /merge-prs tail (owner-approved): src/lib/localDb.ts NEW>800 (799->805, +6 re-exports countFreeProxies + recordFreeProxySyncErrors/clearFreeProxySyncErrors/getFreeProxySyncErrors + FreeProxySyncErrors type for #6909 free-pool relay-repair; re-export-only per Hard Rule #2, not extractable).",
+    "_rebaseline_2026_07_21_8034_compression_exclusions_sidebar": "#8034 (compression exclusions dashboard tab) own growth: sections.ts 796->806 (+10, one new COMPRESSION_CONTEXT_GROUP sidebar item linking /dashboard/compression/exclusions). The file was already 796/800 before this PR (organic growth from prior sidebar entries), so a single new nav item pushed it 6 lines over cap. Freezing at 806 (cannot grow further); the sidebar item array is data, not extractable logic.",
+    "_rebaseline_2026_07_23_8219_cache_ttl_settings_sidebar": "#8219 (@oyi77) own growth: sections.ts 806->813 (+7) — configurable model-catalog cache-TTL settings adds a new sidebar nav entry + its visibility wiring. Sidebar item array is data, not extractable logic; frozen at new size.",
+    "src/shared/constants/sidebarVisibility/sections.ts": 813,
+    "_rebaseline_2026_07_22_8056_headroom_minrows": "#8056 (@RaviTharuma, persist Headroom minRows) own growth: src/lib/db/compression.ts 850->866 (+16 HeadroomConfig+DEFAULT_HEADROOM_CONFIG+normalize/store in get/updateCompressionSettings) and open-sse/services/compression/strategySelector.ts 1054->1060 (+6 merge settings.headroom into stacked stepConfig). Cohesive settings-persistence + stacked-merge wiring at existing chokepoints, frozen at new size.",
+    "_rebaseline_2026_07_22_8081_reasoning_placeholder_guard": "#8081 (@Dingding-leo) own growth: openai-responses.ts 1125->1137 (+12) restructuring the reasoning-placeholder guard so it skips only the empty content block and still emits finish_reason/tool_calls in the same chunk. Cohesive translator wiring; frozen at new size.",
+    "open-sse/services/usage/antigravity.ts": 802,
+    "_rebaseline_2026_07_22_fusion_8013_8098_antigravity": "Fusion of #8013 (backryun, catalog/IDE-CLI-split rewrite) + #8098 (nguyenha935, protocol-fidelity/fail-closed/credits/tool-cloaking): open-sse/services/usage/antigravity.ts NEW 802 (>cap 800, +2 — #8098 credits/tier usage service on #8013's profile-aware headers). Test growth (models-catalog-route 1605->1608, provider-models-route 1752->1757 from #8013 Gemini 3.6 catalog) tracked in testFrozen.",
+    "_rebaseline_2026_07_22_8050_model_lockout_exact_family": "#8050 (@AndrianBalanescu) own growth: accountFallback.ts 1864->1892 (+28) — exact-vs-family model-lockout scoping (getModelLockKey/isModelLocked/clearModelLock/getModelLockoutInfo) so an Antigravity 404 for one bare model no longer hijacks the whole family cooldown. Cohesive lockout logic; frozen at new size."
+  },
   "testCap": 1000,
   "testFrozen": {
     "_rebaseline_2026_06_27_5193_antigravity_test": "#5193 own test growth: oauth-providers-config.test.ts 870->873 (+3: antigravity projectId assertion + 50ms tick for the now fire-and-forget onboarding, matching the no-PKCE/no-openid flow).",
@@ -177,9 +359,9 @@
     "tests/unit/account-fallback-service.test.ts": 1563,
     "tests/unit/batch_api.test.ts": 1324,
     "tests/unit/cc-compatible-provider.test.ts": 1217,
-    "tests/unit/chatcore-translation-paths.test.ts": 2776,
+    "tests/unit/chatcore-translation-paths.test.ts": 2876,
     "tests/unit/chatgpt-web.test.ts": 3148,
-    "tests/unit/combo-routing-engine.test.ts": 3449,
+    "tests/unit/combo-routing-engine.test.ts": 3457,
     "tests/unit/db-migration-runner.test.ts": 1499,
     "tests/unit/deepseek-web.test.ts": 1092,
     "tests/unit/executor-codex.test.ts": 1339,
@@ -189,18 +371,18 @@
     "tests/unit/model-sync-route.test.ts": 1016,
     "tests/unit/models-catalog-route.test.ts": 1636,
     "tests/unit/perplexity-web.test.ts": 1355,
-    "tests/unit/provider-models-route.test.ts": 1784,
+    "tests/unit/provider-models-route.test.ts": 1787,
     "tests/unit/provider-validation-specialty.test.ts": 2985,
     "tests/unit/providers-page-utils.test.ts": 1106,
     "tests/unit/response-sanitizer.test.ts": 1063,
     "tests/unit/route-edge-coverage.test.ts": 1241,
     "tests/unit/search-handler-extended.test.ts": 1071,
-    "tests/unit/sse-auth.test.ts": 1600,
+    "tests/unit/sse-auth.test.ts": 1610,
     "tests/unit/stream-utils.test.ts": 2445,
     "tests/unit/token-refresh-service.test.ts": 1378,
     "tests/unit/translator-openai-responses-req.test.ts": 1194,
-    "tests/unit/translator-openai-to-gemini.test.ts": 1616,
-    "tests/unit/translator-openai-to-kiro.test.ts": 1250,
+    "tests/unit/translator-openai-to-gemini.test.ts": 1619,
+    "tests/unit/translator-openai-to-kiro.test.ts": 1275,
     "tests/unit/translator-resp-gemini-to-openai.test.ts": 1234,
     "tests/unit/usage-service-hardening.test.ts": 1483,
     "tests/unit/vscode-token-routes.test.ts": 1256,
@@ -343,10 +525,10 @@
     "_rebaseline_pr1043_minimax_tts": "Upstream port decolua/9router#1043 (toanalien) own growth: audioSpeech.ts 965->1061 (+96). Adds MiniMax T2A v2 TTS dispatch (handleMinimaxSpeech + hexToBytes helper) — provider entry was already in audioRegistry (format: minimax-tts) but no handler existed, falling through to the OpenAI-compatible default that fails (T2A has custom shape + hex-encoded audio + base_resp envelope). New branch sits next to the other inline provider branches (xiaomi-mimo, coqui, tortoise, aws-polly) — extracting would just create indirection. Covered by tests/unit/minimax-tts-1043.test.ts (3 tests, GREEN: success, base_resp error, invalid-hex).",
     "_rebaseline_pr4592_exclude_exhausted_auto": "Reconcile #4592 already-merged growth: combo.ts 2991->3036 (+45, terminal-status quota-cutoff exclusion in buildAutoCandidates + opt-in gate). Fast-gate PR->release does not run check:file-size.",
     "open-sse/executors/antigravity.ts": 1528,
-    "open-sse/executors/base.ts": 1623,
+    "open-sse/executors/base.ts": 1640,
     "open-sse/executors/chatgpt-web.ts": 3241,
-    "open-sse/executors/codex.ts": 1534,
-    "open-sse/executors/cursor.ts": 1560,
+    "open-sse/executors/codex.ts": 1562,
+    "open-sse/executors/cursor.ts": 1563,
     "open-sse/executors/deepseek-web.ts": 1148,
     "open-sse/executors/grok-web.ts": 1044,
     "open-sse/executors/muse-spark-web.ts": 1405,
@@ -356,18 +538,18 @@
     "open-sse/handlers/search.ts": 1536,
     "open-sse/handlers/videoGeneration.ts": 1063,
     "open-sse/mcp-server/schemas/tools.ts": 1505,
-    "open-sse/mcp-server/server.ts": 1407,
+    "open-sse/mcp-server/server.ts": 1411,
     "open-sse/mcp-server/tools/advancedTools.ts": 1120,
-    "open-sse/services/accountFallback.ts": 1966,
-    "open-sse/services/adobeFireflyClient.ts": 2322,
+    "open-sse/services/accountFallback.ts": 1972,
+    "open-sse/services/adobeFireflyClient.ts": 2385,
     "open-sse/services/claudeCodeCompatible.ts": 1202,
     "open-sse/services/combo.ts": 3648,
     "open-sse/services/compression/strategySelector.ts": 1060,
-    "open-sse/services/rateLimitManager.ts": 1105,
+    "open-sse/services/rateLimitManager.ts": 1167,
     "open-sse/translator/response/openai-responses.ts": 1204,
     "open-sse/utils/cursorAgentProtobuf.ts": 1505,
     "open-sse/utils/stream.ts": 2889,
-    "src/app/(dashboard)/dashboard/HomePageClient.tsx": 1381,
+    "src/app/(dashboard)/dashboard/HomePageClient.tsx": 1388,
     "src/app/(dashboard)/dashboard/analytics/ComboHealthTab.tsx": 1031,
     "src/app/(dashboard)/dashboard/api-manager/ApiManagerPageClient.tsx": 3117,
     "src/app/(dashboard)/dashboard/cache/media/MediaPageClient.tsx": 1067,
@@ -377,7 +559,7 @@
     "src/app/(dashboard)/dashboard/endpoint/EndpointPageClient.tsx": 2615,
     "src/app/(dashboard)/dashboard/health/page.tsx": 1165,
     "src/app/(dashboard)/dashboard/providers/[id]/components/modals/EditConnectionModal.tsx": 1316,
-    "src/app/(dashboard)/dashboard/providers/page.tsx": 1923,
+    "src/app/(dashboard)/dashboard/providers/page.tsx": 1944,
     "src/app/(dashboard)/dashboard/runtime/RuntimePageClient.tsx": 1201,
     "src/app/(dashboard)/dashboard/settings/components/PricingTab.tsx": 1019,
     "src/app/(dashboard)/dashboard/settings/components/ProxyRegistryManager.tsx": 1464,
@@ -386,13 +568,13 @@
     "src/app/(dashboard)/dashboard/settings/components/SystemStorageTab.tsx": 1573,
     "src/app/(dashboard)/dashboard/usage/components/BudgetTab.tsx": 1028,
     "src/app/(dashboard)/dashboard/usage/components/EvalsTab.tsx": 2148,
-    "src/app/(dashboard)/dashboard/usage/components/ProviderLimits/index.tsx": 1109,
+    "src/app/(dashboard)/dashboard/usage/components/ProviderLimits/index.tsx": 1119,
     "src/app/api/providers/[id]/models/route.ts": 2250,
     "src/app/api/v1/models/catalog.ts": 1549,
     "src/lib/tokenHealthCheck.ts": 1021,
     "src/lib/db/apiKeys.ts": 1529,
     "src/lib/db/core.ts": 1637,
-    "src/lib/db/migrationRunner.ts": 1077,
+    "src/lib/db/migrationRunner.ts": 1084,
     "src/lib/db/models.ts": 1097,
     "src/lib/db/providers.ts": 1034,
     "src/lib/memory/retrieval.ts": 1073,
@@ -402,11 +584,13 @@
     "src/shared/components/RequestLoggerV2.tsx": 1629,
     "src/shared/components/analytics/charts.tsx": 1035,
     "src/shared/services/cliRuntime.ts": 1122,
-    "src/sse/handlers/chat.ts": 1845,
+    "src/sse/handlers/chat.ts": 1877,
     "src/sse/services/auth.ts": 2508,
     "tests/unit/account-fallback-service.test.ts": 1572,
     "tests/unit/provider-validation-specialty.test.ts": 2985,
-    "open-sse/executors/hyperagent.ts": 1026
+    "open-sse/executors/hyperagent.ts": 1026,
+    "open-sse/executors/default.ts": 1042,
+    "open-sse/executors/kiro.ts": 1069
   },
   "_rebaseline_2026_07_27_v3849_train2": "Merge-train 2 (7 PRs) — owner-approved 2026-07-27. Single entry: chatCore.ts 4955->5006 (#8595, Responses multi-turn image compaction before the context hard-reject). Genuine irreducible growth at the existing compaction chokepoint in handleChatCore — the PR adds a last-resort retry against the concrete budget plus the estimateFinalInputTokens helper, both wired at the pre-existing call site rather than a new branch. Covered by tests/unit/8560-responses-image-compaction.test.ts (4 tests).",
   "_rebaseline_2026_07_27_v3849_train3": "Merge-train 3 (13 PRs) — owner-approved 2026-07-27. Both entries are genuine irreducible growth at existing chokepoints, not new branches: src/lib/db/apiKeys.ts 1518->1529 (#8805 cx/* ≡ codex/* API-key model permissions); open-sse/handlers/chatCore.ts 5006->5020 (#8806 real response payload into plugin onResponse hooks). Covered by tests/unit/db-apiKeys-crud.test.ts (4 new cases) and the two plugin-hook test files updated in #8806 respectively.",
@@ -415,10 +599,15 @@
   "_rebaseline_2026_07_28_8860_tokenrefresh_projectid": "PR #8860 (fix/antigravity-projectid-centralized) own test growth: tests/unit/token-refresh-service.test.ts 1311->1378 (+67 = 4 cases covering projectId discovery on the tokenRefresh.ts path — the Dashboard/health-check refresh route, which #8842 did not reach since that fixed the executor path). Covered by the same file.",
   "_rebaseline_2026_07_28_8861_xiaomi_token_plan": "PR #8861 (feat/xiaomi-token-plan-protocol-selector) own growth: EditConnectionModal.tsx 1283->1316 (+33 = the per-connection API-protocol selector field) and open-sse/executors/base.ts 1540->1562 (+22 = alternate-format resolution at the existing buildUrl/headers chokepoint). Both are irreducible wiring at existing call sites.",
   "_rebaseline_2026_07_28_8863_firefly_detail_level": "PR #8863 (fix/adobe-firefly-gpt-detail-level-max) own growth: adobeFireflyClient.ts 2317->2322 (+5 = gpt-image detailLevel defaulting to maximal at the existing payload-build site). Covered by tests/unit/adobe-firefly.test.ts.",
+  "_rebaseline_2026_07_28_8870_firefly_ref_cap_timeout": "PR #8870 (fix/adobe-firefly-gpt-ref-cap-timeout) own growth: adobeFireflyClient.ts 2322->2385 (+63 = gpt-image subject-ref hard cap at 2 + adaptive poll timeout budget (base 300s + 60s/ref, max 600s) + defensive .slice on referenceBlobs for gpt/nano/generic families). Fixes live 504s on multi-screenshot listing jobs (Featured Promo / Box Art) where 3–4+ subject refs stall colligo until the old 180s poll budget expires. Helpers adobeFireflyMaxImageRefs/adobeFireflyImageTimeoutMs live next to the existing payload/poll chokepoint (not extractable without splitting the wire recipe mid-PR). Covered by tests/unit/adobe-firefly.test.ts (ref-cap + timeout cases). Structural shrink tracked in #3501.",
   "_rebaseline_2026_07_29_8281_home_quickstart_prefetch": "Release v3.8.49 base-red fix (no PR — captain sweep): src/app/(dashboard)/dashboard/HomePageClient.tsx 1377->1381 (+4). #8292 added prefetch={false} to the sidebar but left /home's five quick-start Links prefetching, so first paint still fired 12 speculative RSC requests — caught by navigation.spec.ts only after the e2e helper bug (APP_ROUTE_PATTERN missing /home) was repaired in the same cycle. Growth is the five prefetch attributes; it was offset first by extracting the repeated className literals (INLINE_LINK x4, DOCS_LINK x1), which collapsed five wrapped <Link> blocks back to one line each — a naive fix measured 1391. Guard: tests/unit/sidebar-prefetch-policy-8281.test.ts.",
   "_rebaseline_2026_08_01_8964_xai_agent_tools": "PR #8964 own growth: chatCore.ts 5020->5034 at the existing native-passthrough chokepoint. Adds xAI Agent Tools passthrough for /v1/responses (xai/xai-oauth/xao): resolve nativeXaiResponsesPassthrough, force openai-responses targetFormat, stamp body marker, and OR into the existing nativeCodexPassthrough sites (web-search bypass + requestEndpointPath). Leaf logic in passthroughHelpers, responsesEndpoint, targetFormat, xai executor, responseSanitizer, usageTracking. Cohesive wiring at the Codex passthrough boundary.",
   "_rebaseline_2026_08_01_8964_response_sanitizer": "PR #8964 own growth: responseSanitizer.ts 1115->1128. Keep cost_in_usd_ticks / server_side_tool_usage(_details) through sanitizeResponsesApiResponse allowlists so native xAI tool responses retain usage.",
   "_rebaseline_2026_08_02_v3850_agentrouter_responses": "Release v3.8.50 AgentRouter/Codex compatibility reconciliation. open-sse/executors/base.ts 1562->1578: #9190 wires AgentRouter's selected Claude/OpenAI/Responses protocol through the existing executor URL, auth, identity-header and fingerprint chokepoints; the reusable alternate resolver remains outside base.ts. open-sse/utils/stream.ts 2887->2889: #9213 evaluates Responses ID and usage normalization independently so response.completed always receives finite usage.total_tokens instead of short-circuiting after an ID rewrite. tests/unit/chatcore-translation-paths.test.ts 2769->2776: #9191 updates the existing Claude-Code bridge assertions for the dynamic AgentRouter wire image. PR #9224 offsets its own chatCore growth by extracting the AgentRouter protocol decisions into chatCore/agentRouterProtocol.ts, leaving chatCore below its frozen ceiling. Covered by agentrouter executor/chatCore protocol tests, chatcore translation-path tests, and responses-commentary-passthrough tests.",
   "_rebaseline_2026_08_05_9323_agentrouter_waf_retry": "PR #9323 (fix(agentrouter): retry on 400 content-blocked + burst guard) own growth: open-sse/executors/base.ts 1578->1623 (check-file-size.mjs conta via split(\"\\n\").length; wc -l ve 1622). As +45 linhas sao o WAF_RETRY_CONFIG + o burst guard via gateOutboundRequest() para o WAF do agentrouter.org, com comentarios explicando o porque de cada mitigacao e cobertos por tests/unit/base-executor-waf-retry.test.ts e tests/unit/wafRateLimit.test.ts. Crescimento funcional legitimo, nao inchaco.",
-  "_rebaseline_2026_08_05_9529_own_growth": "PR #9529 own growth (base release/v3.8.50 medida EXATAMENTE nos frozen antigos, entao o modo base-relative #8522 nao cobre): open-sse/services/rateLimitManager.ts 1060->1105 (+45: helper applyLimiterSettings() que re-arma o heartbeat do reservoir apos updateSettings — fix do bug Bottleneck 2.19.5 que congelava a fila weighted; TDD em tests/unit/ratelimit-reservoir-refresh.test.ts); tests/integration/chat-pipeline.test.ts 1592->1598 (+6: User-Agent do codex derivado de getCodexClientVersion() em vez de literal pinado — teste-irmao alinhado ao contrato); tests/unit/provider-validation-specialty.test.ts 2980->2985 (+5: cobertura NOVA claude-web 429 -> valid:false, alinhamento #9406); open-sse/translator/response/openai-responses.ts 1174->1204 (+30: buildResponsesReasoningSummaryDelta MOVIDA do leaf pureHelpers.ts para o host — a funcao do #9500 muta stream state e violava o contrato do leaf puro; o LOC total do par host+leaf nao cresceu, o pureHelpers encolheu o mesmo tanto). Crescimento por fix de producao + cobertura adicional + realocacao arquitetural, nao inchaco."
+  "_rebaseline_2026_08_05_9529_own_growth": "PR #9529 own growth (base release/v3.8.50 medida EXATAMENTE nos frozen antigos, entao o modo base-relative #8522 nao cobre): open-sse/services/rateLimitManager.ts 1060->1105 (+45: helper applyLimiterSettings() que re-arma o heartbeat do reservoir apos updateSettings — fix do bug Bottleneck 2.19.5 que congelava a fila weighted; TDD em tests/unit/ratelimit-reservoir-refresh.test.ts); tests/integration/chat-pipeline.test.ts 1592->1598 (+6: User-Agent do codex derivado de getCodexClientVersion() em vez de literal pinado — teste-irmao alinhado ao contrato); tests/unit/provider-validation-specialty.test.ts 2980->2985 (+5: cobertura NOVA claude-web 429 -> valid:false, alinhamento #9406); open-sse/translator/response/openai-responses.ts 1174->1204 (+30: buildResponsesReasoningSummaryDelta MOVIDA do leaf pureHelpers.ts para o host — a funcao do #9500 muta stream state e violava o contrato do leaf puro; o LOC total do par host+leaf nao cresceu, o pureHelpers encolheu o mesmo tanto). Crescimento por fix de producao + cobertura adicional + realocacao arquitetural, nao inchaco.",
+  "_rebaseline_2026_08_06_v3850_inherited_drift_reconcile": "Reconciliacao 2026-08-06 do drift ACUMULADO da release/v3.8.50 apos o lote de merges de 08-05/06: 13 arquivos acima do frozen no tip puro 8180b49ce1 (medidos pelo proprio gate). O modo PR base-relative (#8522) deixa PRs inocentes passarem, e os rebaselines individuais dos PRs se perderam nas resolucoes sucessivas de conflito deste hot-file — o drift so aparece no modo absoluto (nightly/local). Crescimentos funcionais dos PRs mergeados: #9024 topology click-nav src/app/(dashboard)/dashboard/HomePageClient.tsx; #9324 OpenRouter enrich src/app/(dashboard)/dashboard/providers/page.tsx; #9329 quota card ordering src/app/(dashboard)/dashboard/usage/components/ProviderLimits/index.tsx; #9193 context-window suffixes src/sse/handlers/chat.ts; #9332 nested Claude server tool ids open-sse/executors/base.ts; #9228 strip orphaned tool outputs open-sse/executors/codex.ts; #9236 nvidia tool-name normalize open-sse/executors/default.ts; #9314 nested tool_call validation open-sse/executors/kiro.ts; #9260 caller identity REST hops open-sse/mcp-server/server.ts; #8934 cache breakpoints tests tests/unit/chatcore-translation-paths.test.ts; #9193 suffix tests tests/unit/combo-routing-engine.test.ts; #9196 reasoning-on-tool-finish tests tests/unit/sse-auth.test.ts; #9163 GPT-5.6 Max reasoning tests tests/unit/translator-openai-to-kiro.test.ts. default.ts e kiro.ts entram no frozen (estavam sem entrada, acima do cap 1000). Atualizacao pos-medicao (a base avancou durante o ciclo do PR): src/sse/handlers/chat.ts 1857->1877 (#9184 affinity EOF evict) e open-sse/executors/default.ts 1027->1042 (#9005 Kimi K3 tool-name backfill).",
+  "_rebaseline_2026_08_06b_v3850_sweepreds_drift": "Segunda reconciliacao de 2026-08-06 (/sweep-reds sobre o tip puro 2ddbbc61a6): 3 arquivos voltaram a passar do frozen apos os merges do mesmo dia, com atribuicao 1:1 por commit. (1) src/app/(dashboard)/dashboard/providers/page.tsx 1928->1944 e (2) open-sse/executors/base.ts 1635->1640, ambos do #9515 (feat(radar): flag-gated signed free-model catalog overlay, commit e7f6b1d130) — o overlay do Radar entra por wiring nos chokepoints ja existentes (a resolucao/verificacao do catalogo assinado mora fora destes dois arquivos); +16 e +5 linhas liquidas nao sao extraiveis sem inventar um leaf por callsite. (3) open-sse/services/accountFallback.ts 1966->1972 do #8704 (commit c4527f97bd), +6 linhas de dados em CREDITS_EXHAUSTED_SIGNALS ('has been exhausted', fixes #8631). src/sse/handlers/chat.ts 1880>1877 tambem estava violando e NAO entra aqui de proposito: e drenado por encolhimento na PR #9598, sem rebaseline. Crescimento proprio DESTA PR: src/lib/db/migrationRunner.ts 1077->1084 (+7) — o guard retroativo em isSchemaAlreadyApplied para os arquivos renumerados 137/138, exigido pela propria mensagem de erro de colisao do runner (ambas as migracoes sao ALTER TABLE ADD COLUMN puro, nao idempotente). Dois `case` + dois `return hasColumn(...)` + 3 linhas de comentario dentro do switch existente; nao extraivel.",
+  "_rebaseline_2026_08_06c_v3850_sweepreds_pr2": "Segunda PR do /sweep-reds (fix/release-v3.8.50-basereds-0806b): tests/unit/provider-models-route.test.ts 1784->1787 (medido pelo gate, que conta split(\"\\n\").length) (+2 apos compressao de comentarios) — alinhamento de contrato forcado por dois merges do dia: #9106 tornou gemini-3.1-pro-high user-callable (a entry do alias entra na lista esperada do teste de discovery-retry, +1 linha de dado + 1 de comentario) e ff012ff420 adicionou onboardUser como bootstrap hop (exclusao no mock, ja comprimida a 1 linha). Nao ha o que encolher sem apagar o comentario que explica o porque.",
+  "_rebaseline_2026_08_07_v3850_sweepreds_pr2_toolnamemap": "tests/unit/translator-openai-to-gemini.test.ts 1616->1619 (+3). O frozen estava EXATAMENTE no tamanho da base, entao qualquer linha nova viola. #9568 (c9a3361e5a) fez buildChangedToolNameMap emitir entradas IDENTIDADE (o Gemini minusculiza nomes de tool nas respostas, entao o tradutor de resposta precisa da chave para mapear de volta), o que passou a incluir `_toolNameMap` no envelope Antigravity de qualquer request com tools. As 3 linhas sao: a chave nova na lista esperada de Object.keys, 1 comentario explicando POR QUE ela aparece (sem ele o proximo leitor tenta remove-la de novo) e 1 assert do CONTEUDO do map — presenca de chave sozinha nao provaria a entrada identidade, que e justamente o comportamento novo. Nao ha o que extrair: e alinhamento de contrato dentro de um teste existente."
 }

config/quality/quality-baseline.json

@@ -82,9 +82,10 @@
       "tightenSlack": 10
     },
     "openapiCoverage.pct": {
-      "value": 38,
+      "value": 39.2,
       "direction": "up",
       "eps": 0.5,
+      "_tighten_2026_08_06_v3850_sweepreds": "38.0 -> 39.2 (aperto EXIGIDO pelo step 'Require-tighten (blocking)', que estava vermelho em ~60 PRs abertas de release/v3.8.50 — base-red herdado, nao defeito das PRs). A cobertura melhorou no ciclo porque as rotas novas entraram documentadas. 39.2 = valor medido pelo CI Quality Ratchet no run 31088889488; o tip puro 2ddbbc61a6 mede 39.3 localmente (npm run check:openapi-coverage: 247/628 rotas), entao 39.2 e o valor conservador dos dois. Aperto = gate mais ESTRITO, nunca mascaramento.",
       "_tighten_2026_07_04_v3844_release": "36.9 -> 39.3 (aperto exigido pelo --require-tighten no PR de release #5925). A cobertura OpenAPI melhorou no ciclo (9 rotas documentadas em 8fb020676 + as rotas novas de #5939/#5817/#6034/#5998 documentadas junto das features). 39.3 = valor medido pelo CI Quality Ratchet no run 28708141003 (tip 00c55afcb).",
       "_rebaseline_2026_06_28_v3839_release": "37.8 -> 36.9 (-0.9, beyond the 0.5 eps). v3.8.39 cycle drift surfaced ONLY on the release PR (the openapi-coverage ratchet does NOT run on PR->release fast-gates). The cycle added API/internal routes (antigravity paste-credentials onboarding, CCR ranged/grep/stats retrieve params, mcp 404 session handling) faster than docs/openapi.yaml coverage; documenting LOCAL_ONLY/internal onboarding routes in the PUBLIC spec would be gaming (same precedent as _rebaseline_2026_06_18_v3828_cycle_close). Measured by CI collect-metrics (run 28317145160) = 36.9. My release-finalize tree touches no routes (only the openapi.yaml version bump). Raising coverage by documenting public routes is tracked as follow-up doc debt.",
       "_rebaseline_2026_06_23_v3834_release": "38.4 -> 37.8 (-0.6, beyond the 0.5 eps so it failed the ratchet). v3.8.34 cycle drift: contributor PRs added API routes (e.g. quota/usage/opencode-go endpoints) faster than openapi.yaml coverage; the openapi-coverage ratchet does NOT run on PR->release fast-gates so it surfaced only on the release PR. Verified my release-finalize working tree touches no routes / openapi paths (only version bump in openapi.yaml). Measured by CI quality:collect (run 28000387577) = 37.8. Raising coverage by documenting the new routes is tracked as follow-up doc debt.",
@@ -109,6 +110,8 @@
       "_rebaseline_2026_06_26_v3837_release": "343->345. v3.8.37 cycle drift surfaced by the release-green pre-flight (the Quality Ratchet does NOT run on PR->release fast-gates, so warnings/complexity accrued unmeasured across this cycle's 76 commits — provider adds DGrid/Pioneer/xAI, headroom proxy lifecycle #4649, ~50 SSE/translator fixes, Engine Combos #5062). Trust-but-verify: this release-finalize working tree touches ONLY CHANGELOG.md, docs/i18n/*/CHANGELOG.md mirrors, and these baselines — 0 production-code change, so all drift is inherited cycle drift (`any` warn-allowed in open-sse/ + tests/). Tighten via --require-tighten next cycle."
     },
     "cognitiveComplexity": {
+      "value": 957,
+      "_rebaseline_2026_07_25_dario_upstream_proxy_selector": "951->957 (+6). Same cycle-drift + own-growth split as the complexity-baseline.json note dated 2026-07-25 (PR #8523, Dario embedded service): cognitive-complexity does not run on PR->release fast-gates, so drift accrues unratcheted. Base upstream/release/v3.8.49 tip measures 956 locally with this PR\u0027s commits removed; this branch measures 957 both locally and on the CI runner. This PR\u0027s own genuine contribution is +1: the new mode-selector conditional rendering (Native/CLIProxyAPI/Dario/Fallback branches plus the fallback-backend picker) in ConnectionRow.tsx. Structural shrink stays tracked in #3501. Tighten via --update next cycle.",
       "value": 1223,
       "_rebaseline_2026_07_25_8470_hyperagent_sticky_thread": "951->957 (+6). PR #8470 (artickc, fix/hyperagent-tool-loop-thread-sticky) pre-green validation. Trust-but-verify: origin/release/v3.8.49 tip alone (pristine, no PR changes) already measures 956 with node scripts/check/check-cognitive-complexity.mjs — i.e. +5 is inherited cycle drift unrelated to this PR (cognitive-complexity does not run on PR->release fast-gates). This PR's OWN growth adds exactly +1: per-file eslint scoped scan (eslint --config eslint.complexity-ratchets.config.mjs open-sse/executors/hyperagent.ts) on base vs PR shows extractMessageText() crossing the threshold for the first time (new sonarjs/cognitive-complexity violation, 26 > 15) from the new Anthropic tool_use/tool_result flattening branches; resolveHyperAgentThreadBinding's existing pre-#8470 violation (16) grows to 21 (still counted once) from the new root-key lookup tier; createHyperAgentThread and execute() are unchanged pre-existing violations. Net repo-wide total = 956 (inherited drift) + 1 (this PR's own new violation) = 957. Full-repo re-measurement of the merged branch was attempted but not completed live due to heavy concurrent devbox load (many other /green-prs sessions running the identical full-repo eslint scan in parallel); the value here is derived from two independently-clean measurements (base-tip full scan + per-file base-vs-PR delta) rather than a third full-repo run. Covered by tests/unit/executor-hyperagent.test.ts (19/19). Tighten via --update next cycle.",
       "_rebaseline_2026_07_25b_v3849_mergetrain_owngrowth": "Owner-approved (chat, 2026-07-25): 956->968 (+12). v3.8.49 /merge-prs 41-PR merge-train aggregate own-growth: measured 968 on the combined boarded tree (tip ac15014ca7) vs 956 on the pristine release tip. The batch's new over-threshold functions come from the pre-screen-flagged complexity-growth set (#8378/#8432/#8476/#8526 etc); each PR is under-ceiling alone, the combined batch adds +12. Same merge-burst class as the notes below; owner chose ceiling-absorb over per-PR extraction. Structural shrink tracked in #3501; tighten via --update next cycle.",
@@ -341,6 +344,12 @@
       "value": 83.18,
       "direction": "up",
       "dedicatedGate": true
+    },
+    "rtlPhysicalClasses": {
+      "value": 1011,
+      "direction": "down",
+      "eps": 0,
+      "_seeded_2026_07_28": "Measured by scripts/check/check-rtl-ratchet.mjs. tests/unit/ui/rtl-logical-classes.test.tsx pins four components and states it is partial (#3541); this bounds the remainder."
     }
   },
   "_coverage_note": "Pisos anti-flake ~2pt abaixo do real do CI mergeado MEDIDO COM os 135 testes religados (run 27247237268: statements 78.4 / lines 78.4 / functions 83.84 / branches 75.73). O religamento da 6A.1 HONESTIFICOU a regua: os ~82.5 anteriores eram inflados porque modulos nunca importados ficavam fora do denominador do c8. Apertar via --require-tighten na Fase 6A (2026-06-16).",

docker-compose.prod.yml

@@ -29,7 +29,7 @@
 services:
   # ── Redis (Rate Limiter Backend) ──────────────────────────────────
   redis:
-    image: redis:8.6.2-alpine
+    image: redis:8.6.5-alpine
     container_name: omniroute-redis-prod
     restart: unless-stopped
     volumes:

docker-compose.yml

@@ -42,6 +42,7 @@ x-common: &common
     - LIVE_WS_HOST=${LIVE_WS_HOST:-0.0.0.0}
     - LIVE_WS_ALLOWED_ORIGINS=${LIVE_WS_ALLOWED_ORIGINS:-http://localhost:20128,http://127.0.0.1:20128}
     - REDIS_URL=${REDIS_URL:-redis://redis:6379}
+    - NODE_OPTIONS=--max-old-space-size=2048
   volumes:
     - ./data:/app/data
   healthcheck:
@@ -54,7 +55,7 @@ x-common: &common
 services:
   # ── Redis (Rate Limiter Backend) ──────────────────────────────────
   redis:
-    image: docker.io/library/redis:7-alpine
+    image: docker.io/library/redis:8.6.5-alpine
     container_name: omniroute-redis
     restart: unless-stopped
     ports:

docker/devin-bridge/Dockerfile

@@ -0,0 +1,57 @@
+FROM node:26.0.0-bookworm-slim
+
+ARG CLAUDE_CODE_VERSION=2.1.220
+ARG DEVIN_CLI_VERSION=3000.2.17
+ARG TARGETARCH
+
+RUN apt-get update \
+  && apt-get install -y --no-install-recommends ca-certificates curl git bash python3 make g++ tini \
+  && rm -rf /var/lib/apt/lists/* \
+  && npm install --global "@anthropic-ai/claude-code@${CLAUDE_CODE_VERSION}"
+
+RUN set -eu; \
+  case "${TARGETARCH}" in \
+    amd64) devin_arch=x86_64-unknown-linux; devin_sha=f0e1e9363afc6ee68c4ef87bab4aeb7ff5cc08a5fa838350ef3ceefdbb2a2be2 ;; \
+    arm64) devin_arch=aarch64-unknown-linux; devin_sha=116dc71ef085a922bc3ff0ea0377d4b26c529a431d58246e36572913e2d25624 ;; \
+    *) echo "Unsupported TARGETARCH=${TARGETARCH}" >&2; exit 1 ;; \
+  esac; \
+  curl -fsSL "https://static.devin.ai/cli/${DEVIN_CLI_VERSION}/devin-${DEVIN_CLI_VERSION}-${devin_arch}.tar.gz" -o /tmp/devin.tar.gz; \
+  echo "${devin_sha}  /tmp/devin.tar.gz" | sha256sum -c -; \
+  tar -xzf /tmp/devin.tar.gz -C /tmp; \
+  install -m 0755 "$(find /tmp -type f -name devin | head -1)" /usr/local/bin/devin; \
+  rm -rf /tmp/devin.tar.gz /tmp/devin-*
+
+RUN groupadd --gid 10001 bridge \
+  && useradd --uid 10001 --gid bridge --create-home --home-dir /home/bridge --shell /bin/bash bridge \
+  && mkdir -p /opt/omniroute /workspace \
+  && chown -R bridge:bridge /opt/omniroute /workspace
+
+WORKDIR /opt/omniroute
+USER bridge
+COPY --chown=bridge:bridge package.json package-lock.json .npmrc ./
+RUN npm ci --ignore-scripts --no-audit --fund=false
+COPY --chown=bridge:bridge . .
+RUN npm rebuild better-sqlite3 || true
+
+ENV HOME=/home/bridge \
+  CLAUDE_CONFIG_DIR=/home/bridge/.claude-devin-isolated \
+  DEVIN_AGENTIC_HOME=/home/bridge \
+  DATA_DIR=/home/bridge/.omniroute-isolated \
+  SQLITE_FILE=/home/bridge/.omniroute-isolated/storage.sqlite \
+  CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1 \
+  DISABLE_TELEMETRY=1 \
+  DISABLE_ERROR_REPORTING=1 \
+  DISABLE_AUTOUPDATER=1 \
+  CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY=1 \
+  NEXT_TELEMETRY_DISABLED=1
+
+RUN mkdir -p /home/bridge/.claude-devin-isolated /home/bridge/.local/share/devin \
+  /home/bridge/.omniroute-isolated
+
+RUN DATA_DIR=/tmp/omniroute-build-data \
+  SQLITE_FILE=/tmp/omniroute-build-data/storage.sqlite \
+  npm run build \
+  && rm -rf /tmp/omniroute-build-data
+
+ENTRYPOINT ["/usr/bin/tini", "--"]
+CMD ["bash"]

docker/devin-bridge/compose.yml

@@ -0,0 +1,218 @@
+name: omniroute-devin-bridge
+
+x-isolated-environment: &isolated-environment
+  HOME: /home/bridge
+  CLAUDE_CONFIG_DIR: /home/bridge/.claude-devin-isolated
+  DEVIN_AGENTIC_HOME: /home/bridge
+  DATA_DIR: /home/bridge/.omniroute-isolated
+  SQLITE_FILE: /home/bridge/.omniroute-isolated/storage.sqlite
+  ANTHROPIC_BASE_URL: http://omniroute:20128
+  ANTHROPIC_AUTH_TOKEN: sk-local-devin-gateway
+  CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC: "1"
+  DISABLE_TELEMETRY: "1"
+  DISABLE_ERROR_REPORTING: "1"
+  DISABLE_AUTOUPDATER: "1"
+  CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY: "1"
+  DEVIN_BRIDGE_MODEL: ${DEVIN_BRIDGE_MODEL:-devin-cli-agentic/swe-1-7}
+  ANTHROPIC_MODEL: ${DEVIN_BRIDGE_MODEL:-devin-cli-agentic/swe-1-7}
+  ANTHROPIC_DEFAULT_SONNET_MODEL: ${DEVIN_BRIDGE_SONNET_MODEL:-devin-cli-agentic/swe-1-7}
+  ANTHROPIC_DEFAULT_OPUS_MODEL: ${DEVIN_BRIDGE_OPUS_MODEL:-devin-cli-agentic/swe-1-7}
+  ANTHROPIC_DEFAULT_HAIKU_MODEL: ${DEVIN_BRIDGE_HAIKU_MODEL:-devin-cli-agentic/swe-1-7}
+  CLAUDE_CODE_SUBAGENT_MODEL: ${DEVIN_BRIDGE_SUBAGENT_MODEL:-devin-cli-agentic/swe-1-7}
+  REQUIRE_API_KEY: "true"
+  OMNIROUTE_API_KEY: sk-local-devin-gateway
+
+x-runtime: &runtime
+  image: omniroute-devin-bridge:local
+  build:
+    context: ../..
+    dockerfile: docker/devin-bridge/Dockerfile
+    args:
+      CLAUDE_CODE_VERSION: 2.1.220
+      DEVIN_CLI_VERSION: 3000.2.17
+  user: "10001:10001"
+  read_only: true
+  tmpfs:
+    - /tmp:rw,noexec,nosuid,nodev,size=256m
+    - /opt/omniroute/.source:rw,nosuid,nodev,size=16m,uid=10001,gid=10001
+  cap_drop: [ALL]
+  security_opt: [no-new-privileges:true]
+  environment: *isolated-environment
+  networks: [bridge-internal]
+
+services:
+  omniroute:
+    <<: *runtime
+    profiles: [offline]
+    hostname: omniroute
+    environment:
+      <<: *isolated-environment
+      CLI_DEVIN_AGENTIC_BIN: /opt/omniroute/docker/devin-bridge/mock-devin.mjs
+      DEVIN_BRIDGE_MOCK_LOG: /evidence/mock-acp.jsonl
+    command: ["npm", "run", "start"]
+    healthcheck:
+      test:
+        [
+          "CMD",
+          "node",
+          "-e",
+          "fetch('http://127.0.0.1:20128/healthz').then(r=>process.exit(r.ok?0:1)).catch(()=>process.exit(1))",
+        ]
+      interval: 2s
+      timeout: 2s
+      retries: 60
+    volumes:
+      - omniroute-offline-data:/home/bridge/.omniroute-isolated
+      - ../../.sandbox/evidence:/evidence
+      - ./mock-devin.mjs:/opt/omniroute/docker/devin-bridge/mock-devin.mjs:ro
+
+  claude:
+    <<: *runtime
+    profiles: [offline]
+    depends_on:
+      omniroute:
+        condition: service_healthy
+      claude-egress-guard:
+        condition: service_healthy
+    working_dir: /workspace
+    command: ["bash", "/opt/omniroute/docker/devin-bridge/run-claude-e2e.sh"]
+    environment:
+      <<: *isolated-environment
+      NODE_USE_ENV_PROXY: "1"
+      HTTP_PROXY: http://claude-egress-guard:8080
+      HTTPS_PROXY: http://claude-egress-guard:8080
+      NO_PROXY: omniroute
+    volumes:
+      - claude-isolated-config:/home/bridge/.claude-devin-isolated
+      - ../../.sandbox/e2e-workspace:/workspace
+      - ../../.sandbox/evidence:/evidence
+      - ./run-claude-e2e.sh:/opt/omniroute/docker/devin-bridge/run-claude-e2e.sh:ro
+
+  contract:
+    <<: *runtime
+    profiles: [offline]
+    depends_on:
+      omniroute:
+        condition: service_healthy
+    command: ["node", "/opt/omniroute/docker/devin-bridge/run-contract.mjs"]
+    volumes:
+      - ./run-contract.mjs:/opt/omniroute/docker/devin-bridge/run-contract.mjs:ro
+
+  claude-egress-guard:
+    image: node:26.0.0-bookworm-slim
+    profiles: [offline, live-devin]
+    user: "10001:10001"
+    read_only: true
+    cap_drop: [ALL]
+    security_opt: [no-new-privileges:true]
+    command: ["node", "/guard/proxy.mjs"]
+    environment:
+      GUARD_LISTEN: 0.0.0.0:8080
+      GUARD_POLICY: deny-all
+      GUARD_LOG: /guard-audit/egress.jsonl
+    healthcheck:
+      test:
+        [
+          "CMD",
+          "node",
+          "-e",
+          "require('net').connect(8080,'127.0.0.1').on('connect',()=>process.exit(0)).on('error',()=>process.exit(1))",
+        ]
+      interval: 1s
+      timeout: 1s
+      retries: 15
+    volumes:
+      - ./network-guard:/guard:ro
+      - ../../.sandbox/guard-audit/claude:/guard-audit
+    networks: [bridge-internal]
+
+  network-guard:
+    image: node:26.0.0-bookworm-slim
+    profiles: [live-devin]
+    user: "10001:10001"
+    read_only: true
+    cap_drop: [ALL]
+    security_opt: [no-new-privileges:true]
+    command: ["node", "/guard/proxy.mjs"]
+    environment:
+      GUARD_LISTEN: 0.0.0.0:8080
+      GUARD_POLICY: devin
+      GUARD_LOG: /guard-audit/egress.jsonl
+    healthcheck:
+      test:
+        [
+          "CMD",
+          "node",
+          "-e",
+          "require('net').connect(8080,'127.0.0.1').on('connect',()=>process.exit(0)).on('error',()=>process.exit(1))",
+        ]
+      interval: 1s
+      timeout: 1s
+      retries: 15
+    volumes:
+      - ./network-guard:/guard:ro
+      - ../../.sandbox/guard-audit/devin:/guard-audit
+    networks: [devin-guard-internal, guard-egress]
+
+  omniroute-live:
+    <<: *runtime
+    profiles: [live-devin]
+    hostname: omniroute
+    depends_on:
+      network-guard:
+        condition: service_healthy
+    environment:
+      <<: *isolated-environment
+      CLI_DEVIN_AGENTIC_BIN: /usr/local/bin/devin
+      DEVIN_BRIDGE_PROXY_URL: http://network-guard:8080
+    networks: [bridge-internal, devin-guard-internal]
+    command: ["npm", "run", "start"]
+    healthcheck:
+      test:
+        [
+          "CMD",
+          "node",
+          "-e",
+          "fetch('http://127.0.0.1:20128/healthz').then(r=>process.exit(r.ok?0:1)).catch(()=>process.exit(1))",
+        ]
+      interval: 2s
+      timeout: 2s
+      retries: 60
+    volumes:
+      - devin-auth:/home/bridge/.local/share/devin
+      - omniroute-live-data:/home/bridge/.omniroute-isolated
+
+  claude-live:
+    <<: *runtime
+    profiles: [live-devin]
+    depends_on:
+      omniroute-live:
+        condition: service_healthy
+      claude-egress-guard:
+        condition: service_healthy
+    working_dir: /workspace
+    command: ["bash", "/opt/omniroute/docker/devin-bridge/run-claude-live-e2e.sh"]
+    environment:
+      <<: *isolated-environment
+      NODE_USE_ENV_PROXY: "1"
+      HTTP_PROXY: http://claude-egress-guard:8080
+      HTTPS_PROXY: http://claude-egress-guard:8080
+      NO_PROXY: omniroute
+    volumes:
+      - claude-isolated-config:/home/bridge/.claude-devin-isolated
+      - ../../.sandbox/live-workspace:/workspace
+      - ../../.sandbox/evidence:/evidence
+      - ./run-claude-live-e2e.sh:/opt/omniroute/docker/devin-bridge/run-claude-live-e2e.sh:ro
+
+networks:
+  bridge-internal:
+    internal: true
+  devin-guard-internal:
+    internal: true
+  guard-egress: {}
+
+volumes:
+  claude-isolated-config: {}
+  devin-auth: {}
+  omniroute-offline-data: {}
+  omniroute-live-data: {}

docker/devin-bridge/mock-devin.mjs

@@ -0,0 +1,229 @@
+#!/usr/bin/env node
+import fs from "node:fs";
+import readline from "node:readline";
+
+if (
+  process.argv[2] !== "acp" ||
+  process.argv[3] !== "--agent-type" ||
+  process.argv[4] !== "summarizer" ||
+  process.argv.length !== 5
+) {
+  process.exit(64);
+}
+
+const logFile = process.env.DEVIN_BRIDGE_MOCK_LOG || "/evidence/mock-acp.jsonl";
+const rl = readline.createInterface({ input: process.stdin });
+const send = (value) => process.stdout.write(`${JSON.stringify(value)}\n`);
+const log = (value) => fs.appendFileSync(logFile, `${JSON.stringify(value)}\n`);
+
+const actions = [
+  {
+    name: "Skill",
+    arguments: { skill: "bridge-proof" },
+  },
+  {
+    name: "Bash",
+    arguments: {
+      command: "find . -maxdepth 2 -type f -print",
+      description: "Locate the fixture files",
+    },
+  },
+  {
+    name: "Read",
+    arguments: { file_path: "/workspace/math.js" },
+  },
+  {
+    name: "Edit",
+    arguments: {
+      file_path: "/workspace/math.js",
+      old_string: "return a - b;",
+      new_string: "return a * b;",
+    },
+  },
+  {
+    name: "Bash",
+    arguments: { command: "npm test", description: "Run the fixture tests" },
+  },
+  {
+    name: "Edit",
+    arguments: {
+      file_path: "/workspace/math.js",
+      old_string: "return a * b;",
+      new_string: "return a + b;",
+    },
+  },
+  {
+    name: "Bash",
+    arguments: { command: "npm test", description: "Confirm the corrected fixture" },
+  },
+];
+
+rl.on("line", (line) => {
+  const message = JSON.parse(line);
+  if (message.method === "initialize") {
+    if (message.params?.protocolVersion !== 1) {
+      send({ jsonrpc: "2.0", id: message.id, error: { code: -32602, message: "ACP v1 required" } });
+      return;
+    }
+    send({ jsonrpc: "2.0", id: message.id, result: { protocolVersion: 1 } });
+  } else if (message.method === "session/new") {
+    if (message.params?.cwd !== "/home/bridge" || !Array.isArray(message.params?.mcpServers)) {
+      send({ jsonrpc: "2.0", id: message.id, error: { code: -32602, message: "unsafe session" } });
+      return;
+    }
+    send({
+      jsonrpc: "2.0",
+      id: message.id,
+      result: { sessionId: "offline" },
+    });
+  } else if (message.method === "session/set_config_option") {
+    send({
+      jsonrpc: "2.0",
+      id: message.id,
+      error: { code: -32602, message: "summarizer mode must not be mutated" },
+    });
+  } else if (message.method === "session/prompt") {
+    const prompt = String(message.params?.prompt?.[0]?.text || "");
+    if (!prompt.includes("[Devin Summarizer Bridge]") || !prompt.includes("[Execution Trace]")) {
+      send({
+        jsonrpc: "2.0",
+        id: message.id,
+        error: { code: -32602, message: "summarizer bridge framing required" },
+      });
+      return;
+    }
+    if (prompt.includes("CONTRACT_AFTER_TOOL")) {
+      log({ provider: "devin-cli-agentic", scenario: "after-tool" });
+      send({
+        jsonrpc: "2.0",
+        method: "session/update",
+        params: {
+          sessionId: "offline",
+          update: {
+            sessionUpdate: "agent_message_chunk",
+            content: { type: "text", text: "contract continued" },
+          },
+        },
+      });
+      send({ jsonrpc: "2.0", id: message.id, result: { stopReason: "end_turn" } });
+      return;
+    }
+    if (prompt.includes("CONTRACT_EXIT")) {
+      log({ provider: "devin-cli-agentic", scenario: "exit" });
+      process.exit(7);
+    }
+    if (prompt.includes("CONTRACT_ERROR")) {
+      log({ provider: "devin-cli-agentic", scenario: "error" });
+      send({
+        jsonrpc: "2.0",
+        id: message.id,
+        error: { code: -32000, message: "deterministic upstream failure" },
+      });
+      return;
+    }
+    if (prompt.includes("CONTRACT_TEXT")) {
+      log({ provider: "devin-cli-agentic", scenario: "text" });
+      send({
+        jsonrpc: "2.0",
+        method: "session/update",
+        params: {
+          sessionId: "offline",
+          update: {
+            sessionUpdate: "agent_message_chunk",
+            content: { type: "text", text: "contract text" },
+          },
+        },
+      });
+      send({ jsonrpc: "2.0", id: message.id, result: { stopReason: "end_turn" } });
+      return;
+    }
+    if (prompt.includes("CONTRACT_NARRATIVE_REPAIR")) {
+      const isRepair = prompt.includes("[Single Repair Attempt]");
+      log({
+        provider: "devin-cli-agentic",
+        scenario: "narrative-repair",
+        stage: isRepair ? "repair" : "initial",
+      });
+      send({
+        jsonrpc: "2.0",
+        method: "session/update",
+        params: {
+          sessionId: "offline",
+          update: {
+            sessionUpdate: "agent_message_chunk",
+            content: {
+              type: "text",
+              text: isRepair
+                ? '<tool>{"name":"Read","arguments":{"file_path":"/workspace/math.js"}}</tool>'
+                : "I'll start by reading the math.js file, then run the tests.",
+            },
+          },
+        },
+      });
+      send({ jsonrpc: "2.0", id: message.id, result: { stopReason: "end_turn" } });
+      return;
+    }
+    if (prompt.includes("CONTRACT_TOOL")) {
+      log({ provider: "devin-cli-agentic", scenario: "tool" });
+      send({
+        jsonrpc: "2.0",
+        method: "session/update",
+        params: {
+          sessionId: "offline",
+          update: {
+            sessionUpdate: "agent_message_chunk",
+            content: {
+              type: "text",
+              text: '<tool>{"name":"Read","arguments":{"file_path":"/workspace/math.js"}}</tool>',
+            },
+          },
+        },
+      });
+      send({ jsonrpc: "2.0", id: message.id, result: { stopReason: "end_turn" } });
+      return;
+    }
+    const resultCount = (prompt.match(/\[Tool Result\]/g) || []).length;
+    if (!prompt.includes("CLAUDE_MD_BRIDGE_ACTIVE") || !prompt.includes("COMMAND_BRIDGE_ACTIVE")) {
+      send({
+        jsonrpc: "2.0",
+        id: message.id,
+        error: { code: -32602, message: "Claude project context missing" },
+      });
+      return;
+    }
+
+    const action = actions[resultCount];
+    const text = action
+      ? `<tool>${JSON.stringify(action)}</tool>`
+      : "BRIDGE_E2E_COMPLETE CLAUDE_MD_BRIDGE_ACTIVE SKILL_BRIDGE_ACTIVE COMMAND_BRIDGE_ACTIVE";
+    if (!action && !prompt.includes("SKILL_BRIDGE_ACTIVE")) {
+      send({
+        jsonrpc: "2.0",
+        id: message.id,
+        error: { code: -32602, message: "Skill result missing" },
+      });
+      return;
+    }
+    log({
+      provider: "devin-cli-agentic",
+      model: message.params?.model || "swe-1-7",
+      resultCount,
+      action: action?.name || "final",
+    });
+    const midpoint = Math.max(1, Math.floor(text.length / 2));
+    for (const chunk of [text.slice(0, midpoint), text.slice(midpoint)]) {
+      send({
+        jsonrpc: "2.0",
+        method: "session/update",
+        params: {
+          sessionId: "offline",
+          update: {
+            sessionUpdate: "agent_message_chunk",
+            content: { type: "text", text: chunk },
+          },
+        },
+      });
+    }
+    send({ jsonrpc: "2.0", id: message.id, result: { stopReason: "end_turn" } });
+  }
+});

docker/devin-bridge/network-guard/policy.mjs

@@ -0,0 +1,130 @@
+export const DEVIN_ALLOWED_SUFFIXES = Object.freeze([".devin.ai", ".cognition.ai"]);
+export const DEVIN_ALLOWED_EXACT_HOSTS = Object.freeze([
+  "server.codeium.com",
+  "unleash.codeium.com",
+]);
+
+function normalizeHostname(hostname) {
+  return String(hostname || "")
+    .trim()
+    .toLowerCase()
+    .replace(/\.$/, "");
+}
+
+export function isAllowedGuardHostname(hostname, policy = "deny-all") {
+  if (policy !== "devin") return false;
+  const value = normalizeHostname(hostname);
+  if (!value) return false;
+  if (DEVIN_ALLOWED_EXACT_HOSTS.includes(value)) return true;
+  return DEVIN_ALLOWED_SUFFIXES.some(
+    (suffix) => value === suffix.slice(1) || value.endsWith(suffix)
+  );
+}
+
+const HOP_BY_HOP_HEADERS = new Set([
+  "connection",
+  "keep-alive",
+  "proxy-authenticate",
+  "proxy-authorization",
+  "proxy-connection",
+  "te",
+  "trailer",
+  "transfer-encoding",
+  "upgrade",
+]);
+
+export function sanitizeForwardHeaders(headers, target) {
+  const connectionTokens = String(headers.connection || "")
+    .split(",")
+    .map((value) => value.trim().toLowerCase())
+    .filter(Boolean);
+  const blocked = new Set([...HOP_BY_HOP_HEADERS, ...connectionTokens]);
+  const sanitized = {};
+  for (const [name, value] of Object.entries(headers)) {
+    if (value === undefined || blocked.has(name.toLowerCase()) || name.toLowerCase() === "host") {
+      continue;
+    }
+    sanitized[name] = value;
+  }
+  sanitized.host = target.host;
+  return sanitized;
+}
+
+export function parseConnectAuthority(authority) {
+  const value = String(authority || "");
+  const match = value.match(/^(?:\[([^\]]+)\]|([^:]+)):(\d+)$/);
+  if (!match) return null;
+  const hostname = normalizeHostname(match[1] || match[2]);
+  const port = Number(match[3]);
+  if (!hostname || port !== 443) return null;
+  return { hostname, port };
+}
+
+function readUint24(buffer, offset) {
+  return (buffer[offset] << 16) | (buffer[offset + 1] << 8) | buffer[offset + 2];
+}
+
+export function parseTlsClientHelloSni(buffer) {
+  if (!Buffer.isBuffer(buffer)) return { status: "invalid", reason: "not_buffer" };
+  let offset = 0;
+  const handshakeParts = [];
+  while (offset < buffer.length) {
+    if (buffer.length - offset < 5) return { status: "need-more" };
+    if (buffer[offset] !== 22) return { status: "invalid", reason: "not_handshake_record" };
+    const recordLength = buffer.readUInt16BE(offset + 3);
+    if (recordLength <= 0 || recordLength > 18432) {
+      return { status: "invalid", reason: "invalid_record_length" };
+    }
+    if (buffer.length - offset - 5 < recordLength) return { status: "need-more" };
+    handshakeParts.push(buffer.subarray(offset + 5, offset + 5 + recordLength));
+    offset += 5 + recordLength;
+  }
+  const handshake = Buffer.concat(handshakeParts);
+  if (handshake.length < 4) return { status: "need-more" };
+  if (handshake[0] !== 1) return { status: "invalid", reason: "not_client_hello" };
+  const helloLength = readUint24(handshake, 1);
+  if (helloLength > 65531) return { status: "invalid", reason: "client_hello_too_large" };
+  if (handshake.length - 4 < helloLength) return { status: "need-more" };
+  const hello = handshake.subarray(4, 4 + helloLength);
+  let cursor = 34;
+  if (hello.length < cursor + 1) return { status: "invalid", reason: "truncated_hello" };
+  const sessionLength = hello[cursor++];
+  cursor += sessionLength;
+  if (hello.length < cursor + 2) return { status: "invalid", reason: "truncated_ciphers" };
+  const cipherLength = hello.readUInt16BE(cursor);
+  cursor += 2 + cipherLength;
+  if (hello.length < cursor + 1) return { status: "invalid", reason: "truncated_compression" };
+  const compressionLength = hello[cursor++];
+  cursor += compressionLength;
+  if (hello.length < cursor + 2) return { status: "invalid", reason: "missing_extensions" };
+  const extensionsLength = hello.readUInt16BE(cursor);
+  cursor += 2;
+  const extensionsEnd = cursor + extensionsLength;
+  if (extensionsEnd > hello.length) return { status: "invalid", reason: "truncated_extensions" };
+  while (cursor < extensionsEnd) {
+    if (extensionsEnd - cursor < 4) return { status: "invalid", reason: "truncated_extension" };
+    const type = hello.readUInt16BE(cursor);
+    const length = hello.readUInt16BE(cursor + 2);
+    cursor += 4;
+    if (cursor + length > extensionsEnd) {
+      return { status: "invalid", reason: "invalid_extension_length" };
+    }
+    if (type === 0) {
+      const data = hello.subarray(cursor, cursor + length);
+      if (data.length < 5 || data.readUInt16BE(0) !== data.length - 2 || data[2] !== 0) {
+        return { status: "invalid", reason: "invalid_server_name" };
+      }
+      const nameLength = data.readUInt16BE(3);
+      if (nameLength !== data.length - 5) {
+        return { status: "invalid", reason: "invalid_server_name_length" };
+      }
+      const serverName = normalizeHostname(data.subarray(5).toString("ascii"));
+      if (!/^[a-z0-9.-]+$/.test(serverName)) {
+        return { status: "invalid", reason: "invalid_server_name_value" };
+      }
+      return { status: "ok", serverName };
+    }
+    cursor += length;
+  }
+  return { status: "invalid", reason: "missing_sni" };
+}

docker/devin-bridge/network-guard/proxy.mjs

@@ -0,0 +1,136 @@
+import fs from "node:fs";
+import http from "node:http";
+import net from "node:net";
+import { pathToFileURL } from "node:url";
+
+import {
+  isAllowedGuardHostname,
+  parseConnectAuthority,
+  parseTlsClientHelloSni,
+  sanitizeForwardHeaders,
+} from "./policy.mjs";
+
+const MAX_CLIENT_HELLO_BYTES = 64 * 1024;
+const CLIENT_HELLO_TIMEOUT_MS = 3000;
+
+export function createGuardProxy({
+  policy = "deny-all",
+  logPath = "/tmp/egress.jsonl",
+  allowHostname = (hostname) => isAllowedGuardHostname(hostname, policy),
+  connectSocket = (port, hostname, onConnect) => net.connect(port, hostname, onConnect),
+} = {}) {
+  if (!new Set(["deny-all", "devin"]).has(policy)) {
+    throw new Error(`Unknown network guard policy: ${policy}`);
+  }
+
+  function audit(hostname, decision, reason) {
+    fs.appendFileSync(
+      logPath,
+      `${JSON.stringify({ at: new Date().toISOString(), hostname, decision, reason })}\n`
+    );
+  }
+
+  const server = http.createServer((req, res) => {
+    let target;
+    try {
+      target = new URL(req.url);
+    } catch {
+      res.writeHead(400).end("invalid proxy target\n");
+      return;
+    }
+    if (target.protocol !== "http:" || target.username || target.password) {
+      audit(target.hostname, "deny", "invalid_http_target");
+      res.writeHead(403).end("egress denied\n");
+      return;
+    }
+    if (!allowHostname(target.hostname)) {
+      audit(target.hostname, "deny", "host_policy");
+      res.writeHead(403).end("egress denied\n");
+      return;
+    }
+    audit(target.hostname, "allow", "host_policy");
+    const upstream = http.request(
+      target,
+      {
+        method: req.method,
+        headers: sanitizeForwardHeaders(req.headers, target),
+      },
+      (reply) => {
+        res.writeHead(reply.statusCode || 502, reply.headers);
+        reply.pipe(res);
+      }
+    );
+    req.pipe(upstream);
+    upstream.on("error", () => res.writeHead(502).end("upstream error\n"));
+  });
+
+  server.on("connect", (req, client, head) => {
+    const authority = parseConnectAuthority(req.url);
+    if (!authority) {
+      audit(req.url, "deny", "invalid_connect_authority");
+      client.end("HTTP/1.1 403 Forbidden\r\n\r\n");
+      return;
+    }
+    const { hostname, port } = authority;
+    if (!allowHostname(hostname)) {
+      audit(hostname, "deny", "host_policy");
+      client.end("HTTP/1.1 403 Forbidden\r\n\r\n");
+      return;
+    }
+
+    let buffer = Buffer.from(head);
+    let settled = false;
+    const timer = setTimeout(() => fail("client_hello_timeout"), CLIENT_HELLO_TIMEOUT_MS);
+    timer.unref?.();
+
+    const cleanup = () => {
+      clearTimeout(timer);
+      client.removeListener("data", onData);
+    };
+    const fail = (reason) => {
+      if (settled) return;
+      settled = true;
+      cleanup();
+      audit(hostname, "deny", reason);
+      client.destroy();
+    };
+    const inspect = () => {
+      if (buffer.length > MAX_CLIENT_HELLO_BYTES) return fail("client_hello_too_large");
+      const parsed = parseTlsClientHelloSni(buffer);
+      if (parsed.status === "need-more") return;
+      if (parsed.status !== "ok") return fail(parsed.reason || "invalid_client_hello");
+      if (parsed.serverName !== hostname) return fail("sni_mismatch");
+      settled = true;
+      cleanup();
+      client.pause();
+      const upstream = connectSocket(port, hostname, () => {
+        audit(hostname, "allow", "sni_match");
+        if (buffer.length) upstream.write(buffer);
+        upstream.pipe(client);
+        client.pipe(upstream);
+        client.resume();
+      });
+      upstream.on("error", () => client.destroy());
+    };
+    const onData = (chunk) => {
+      buffer = Buffer.concat([buffer, chunk]);
+      inspect();
+    };
+
+    client.write("HTTP/1.1 200 Connection Established\r\n\r\n");
+    client.on("data", onData);
+    if (buffer.length) inspect();
+    client.resume();
+  });
+
+  return server;
+}
+
+if (process.argv[1] && pathToFileURL(process.argv[1]).href === import.meta.url) {
+  const [host, portText] = (process.env.GUARD_LISTEN || "0.0.0.0:8080").split(":");
+  const server = createGuardProxy({
+    policy: process.env.GUARD_POLICY || "deny-all",
+    logPath: process.env.GUARD_LOG || "/tmp/egress.jsonl",
+  });
+  server.listen(Number(portText), host);
+}

docker/devin-bridge/run-claude-e2e.sh

@@ -0,0 +1,27 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+unset ANTHROPIC_API_KEY CLAUDE_CODE_OAUTH_TOKEN ANTHROPIC_BEDROCK_BASE_URL ANTHROPIC_VERTEX_BASE_URL
+unset CLAUDE_CODE_USE_BEDROCK CLAUDE_CODE_USE_VERTEX CLAUDE_CODE_USE_FOUNDRY
+
+set -o pipefail
+check() {
+  "$@"
+  printf 'E2E check passed: %s\n' "$*"
+}
+
+claude -p --output-format stream-json --verbose --max-turns 12 \
+  --permission-mode bypassPermissions \
+  "/bridge-check" | tee /evidence/claude-stream.jsonl
+
+if grep -Eqi 'log[ -]?in|authenticate.*anthropic|claude\.ai' /evidence/claude-stream.jsonl; then
+  echo "Claude Code requested forbidden authentication" >&2
+  exit 1
+fi
+check grep -q 'return a + b;' /workspace/math.js
+npm test
+check grep -q 'Skill' /workspace/.e2e-hook.log
+check grep -q 'Read' /workspace/.e2e-hook.log
+check grep -q 'Edit' /workspace/.e2e-hook.log
+check grep -q 'Bash' /workspace/.e2e-hook.log
+check grep -q 'BRIDGE_E2E_COMPLETE' /evidence/claude-stream.jsonl

docker/devin-bridge/run-claude-live-e2e.sh

@@ -0,0 +1,52 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+unset ANTHROPIC_API_KEY CLAUDE_CODE_OAUTH_TOKEN ANTHROPIC_BEDROCK_BASE_URL ANTHROPIC_VERTEX_BASE_URL
+unset CLAUDE_CODE_USE_BEDROCK CLAUDE_CODE_USE_VERTEX CLAUDE_CODE_USE_FOUNDRY
+
+bridge_system_prompt="You are a coding agent inside Claude Code. Use only the client-owned tools supplied in the request. Never execute or request a Devin-owned tool. When work requires a tool, select the appropriate client tool and wait for its result before continuing."
+scenario_cooldown_seconds="${DEVIN_BRIDGE_LIVE_SCENARIO_COOLDOWN_SECONDS:-15}"
+
+run_scenario() {
+  local evidence_file="$1"
+  local prompt="$2"
+  claude -p --output-format stream-json --verbose --max-turns 12 \
+    --tools Read,Edit,Bash \
+    --system-prompt "$bridge_system_prompt" \
+    --permission-mode bypassPermissions "$prompt" | tee "$evidence_file"
+  if grep -Eqi 'log[ -]?in|authenticate.*anthropic|claude\.ai' "$evidence_file"; then
+    echo "Claude Code requested forbidden authentication" >&2
+    exit 1
+  fi
+}
+
+validate_scenario() {
+  local evidence_file="$1"
+  local marker="$2"
+  local required_tools="$3"
+  local require_npm_test="$4"
+  local required_slash_command="${5:-}"
+  local required_skill="${6:-}"
+  local accept_explicit_completion="${7:-false}"
+  node /opt/omniroute/scripts/devin-bridge/validate-claude-evidence.mjs \
+    "$evidence_file" "$marker" "$required_tools" "$require_npm_test" \
+    "$required_slash_command" "$required_skill" "$accept_explicit_completion"
+}
+
+run_scenario /evidence/live-analysis.jsonl \
+  "Read /workspace/CLAUDE.md, /workspace/math.js, and /workspace/math.test.js directly without searching or editing. Explain the defect, then end with LIVE_ANALYSIS_COMPLETE."
+validate_scenario /evidence/live-analysis.jsonl LIVE_ANALYSIS_COMPLETE Read false
+sleep "$scenario_cooldown_seconds"
+
+run_scenario /evidence/live-fix.jsonl \
+  "Use Edit now to replace 'return a - b;' with 'return a + b;' in /workspace/math.js. Then use Bash to run npm test. Do not summarize before npm test succeeds. End with LIVE_FIX_COMPLETE only after the test passes."
+grep -q 'return a + b;' /workspace/math.js
+npm test
+validate_scenario /evidence/live-fix.jsonl LIVE_FIX_COMPLETE Edit,Bash true
+sleep "$scenario_cooldown_seconds"
+
+run_scenario /evidence/live-command.jsonl "/bridge-check"
+validate_scenario /evidence/live-command.jsonl BRIDGE_E2E_COMPLETE Bash true \
+  bridge-check bridge-proof true
+
+printf 'PASS: three live Devin-backed Claude Code scenarios completed\n'

docker/devin-bridge/run-contract.mjs

@@ -0,0 +1,135 @@
+#!/usr/bin/env node
+import assert from "node:assert/strict";
+
+const endpoint = "http://omniroute:20128/v1/messages";
+const headers = {
+  "anthropic-version": "2023-06-01",
+  "content-type": "application/json",
+  "x-api-key": "sk-local-devin-gateway",
+};
+const model = process.env.DEVIN_BRIDGE_MODEL || "devin-cli-agentic/swe-1-7";
+
+async function request(prompt, extra = {}) {
+  return fetch(endpoint, {
+    method: "POST",
+    headers,
+    body: JSON.stringify({
+      model,
+      max_tokens: 256,
+      messages: [{ role: "user", content: prompt }],
+      ...extra,
+    }),
+  });
+}
+
+const textReply = await request("CONTRACT_TEXT");
+assert.equal(textReply.status, 200);
+assert.match(textReply.headers.get("content-type") || "", /application\/json/);
+const textBody = await textReply.json();
+assert.equal(textBody.type, "message");
+assert.equal(textBody.role, "assistant");
+assert.equal(textBody.stop_reason, "end_turn");
+assert.deepEqual(textBody.content, [{ type: "text", text: "contract text" }]);
+
+const toolReply = await request("CONTRACT_TOOL", {
+  stream: true,
+  tools: [
+    {
+      name: "Read",
+      description: "Read a file",
+      input_schema: {
+        type: "object",
+        properties: { file_path: { type: "string" } },
+        required: ["file_path"],
+        additionalProperties: false,
+      },
+    },
+  ],
+});
+assert.equal(toolReply.status, 200);
+assert.match(toolReply.headers.get("content-type") || "", /text\/event-stream/);
+const toolStream = await toolReply.text();
+const eventNames = toolStream
+  .split("\n")
+  .filter((line) => line.startsWith("event: "))
+  .map((line) => line.slice(7));
+assert.deepEqual(eventNames, [
+  "message_start",
+  "content_block_start",
+  "content_block_delta",
+  "content_block_stop",
+  "message_delta",
+  "message_stop",
+]);
+const toolEvents = toolStream
+  .split("\n")
+  .filter((line) => line.startsWith("data: "))
+  .map((line) => JSON.parse(line.slice(6)));
+const toolUse = toolEvents.find((event) => event.type === "content_block_start")?.content_block;
+assert.equal(toolUse?.type, "tool_use");
+assert.equal(toolUse?.name, "Read");
+assert.match(toolUse?.id || "", /^tool_devin_/);
+
+const repairedNarrativeReply = await request("CONTRACT_NARRATIVE_REPAIR", {
+  tools: [
+    {
+      name: "Read",
+      description: "Read a file",
+      input_schema: {
+        type: "object",
+        properties: { file_path: { type: "string" } },
+        required: ["file_path"],
+        additionalProperties: false,
+      },
+    },
+  ],
+});
+assert.equal(repairedNarrativeReply.status, 200);
+const repairedNarrativeBody = await repairedNarrativeReply.json();
+assert.equal(repairedNarrativeBody.stop_reason, "tool_use");
+assert.equal(repairedNarrativeBody.content?.[0]?.type, "tool_use");
+assert.equal(repairedNarrativeBody.content?.[0]?.name, "Read");
+
+const continuationReply = await fetch(endpoint, {
+  method: "POST",
+  headers,
+  body: JSON.stringify({
+    model,
+    max_tokens: 256,
+    tools: [
+      {
+        name: "Read",
+        description: "Read a file",
+        input_schema: { type: "object", properties: {}, additionalProperties: true },
+      },
+    ],
+    messages: [
+      { role: "user", content: "CONTRACT_TOOL" },
+      { role: "assistant", content: [toolUse] },
+      {
+        role: "user",
+        content: [
+          {
+            type: "tool_result",
+            tool_use_id: toolUse.id,
+            content: "CONTRACT_AFTER_TOOL",
+          },
+        ],
+      },
+    ],
+  }),
+});
+assert.equal(continuationReply.status, 200);
+const continuationBody = await continuationReply.json();
+assert.equal(continuationBody.stop_reason, "end_turn");
+assert.deepEqual(continuationBody.content, [{ type: "text", text: "contract continued" }]);
+
+for (const marker of ["CONTRACT_ERROR", "CONTRACT_EXIT"]) {
+  const failedReply = await request(marker);
+  assert.equal(failedReply.status, 502);
+  const failedBody = await failedReply.json();
+  assert.equal(failedBody.error?.type, "server_error");
+  assert.doesNotMatch(JSON.stringify(failedBody), /stack|anthropic|openai/i);
+}
+
+console.log("PASS: Anthropic Messages wire contracts and fail-closed errors passed");

docs/DEVIN_CLAUDE_BRIDGE.md

@@ -0,0 +1,181 @@
+# Devin Claude Bridge
+
+`devin-cli-agentic` lets the real Claude Code runtime use OmniRoute's local Anthropic
+Messages endpoint while the official Devin CLI supplies model responses over ACP stdio. It
+does not modify the existing Anthropic, Claude OAuth, Claude Web, or `devin-cli` providers.
+
+> **Current status: offline and live validated.** The pinned Claude Code `2.1.220` completed
+> three isolated scenarios through Devin CLI `3000.2.17` and model
+> `swe-1-7-lightning`. The final live run proved client-owned `Read`, `Edit`, and `Bash`
+> turns, successful `npm test` results, project command and skill discovery, Devin-only
+> routing, and zero Claude egress.
+
+## Architecture
+
+```text
+Claude Code 2.1.220 (isolated non-root Linux container)
+  -> http://omniroute:20128/v1/messages
+  -> devin-cli-agentic (Claude-format, no-auth provider)
+  -> devin acp --agent-type summarizer (official ACP stdio, no Devin tools)
+  -> Devin account in the dedicated devin-auth volume
+```
+
+The official CLI's default ACP agent can execute its own tools, so this bridge does not use
+it. It starts the fixed `summarizer` ACP agent, whose official CLI mode has no tools, and
+frames the serialized Anthropic request as an execution trace. When another Claude-owned
+action is needed, the response must contain exactly one client tool envelope. Any ACP
+`tool_call` or `tool_call_update` is rejected before a response can be reported as
+successful.
+
+The serializer in `open-sse/executors/devin-agentic/serializer.ts` preserves `system`,
+`text`, `tool_use`, `tool_result`, `thinking`, `redacted_thinking`, `tool_choice`, and the
+tools supplied by Claude Code. Images and unknown blocks fail explicitly. Large tool results
+use a visible truncation marker.
+
+The parser accepts one standalone `<tool>{...}</tool>` envelope per model turn. It checks
+the name against the request's tool list, validates arguments against that tool's JSON
+Schema, rejects mixed narrative/actions, and permits one bounded repair. Claude Code then
+executes the resulting Anthropic `tool_use` locally and sends the `tool_result` back through
+OmniRoute.
+
+## Isolation and threat model
+
+The host's Claude installation, account, and configuration are out of scope and treated as
+forbidden. The Compose services:
+
+- run as UID/GID `10001:10001`, with a read-only root filesystem, dropped capabilities, and
+  `no-new-privileges`;
+- use a private `/home/bridge`, a dedicated Claude config volume, isolated OmniRoute data,
+  and a separate `devin-auth` volume;
+- mount only disposable `.sandbox` workspaces/evidence;
+- do not mount the host home, Keychain, SSH, cloud credentials, or Docker socket;
+- construct explicit environments and remove Anthropic API/OAuth/routing variables;
+- direct Claude Code inference only to `http://omniroute:20128` with a local-only key.
+
+The offline profile uses an internal network. In the live profile, OmniRoute reaches the
+official Devin endpoints only through `network-guard`; unrelated destinations are denied.
+Claude Code has a separate deny-all egress guard and can reach only the local OmniRoute
+service through `NO_PROXY`. Guard audit files are mounted only by their guard process. The
+scripts verify file ownership, mode, link count, and every decision before exporting
+token-free evidence.
+
+Run the isolation proof independently:
+
+```bash
+./scripts/devin-bridge/verify-anthropic-isolation
+```
+
+It validates topology, named mounts, non-root/read-only settings, explicit local routing,
+absence of sensitive environment variables, absence of the Docker socket, blocked access to
+`api.anthropic.com` and `claude.ai`, Devin-only provider selection, and explicit failure when
+the ACP backend is unavailable.
+
+## First-time setup and normal use
+
+Build the pinned image:
+
+```bash
+./scripts/devin-bridge/build
+```
+
+Authenticate only the isolated Devin volume:
+
+```bash
+ENABLE_LIVE_DEVIN_TESTS=1 ./scripts/devin-bridge/login-devin
+```
+
+The login command uses the official manual-token flow intended for remote/container
+environments. The value is entered directly into the CLI prompt; it is not passed as a
+process argument, written to Git, or copied from the host.
+
+Launch the isolated Claude Code runtime:
+
+```bash
+./scripts/devin-bridge/launch
+```
+
+`launch` rechecks isolation, Devin authentication, and model discovery before starting the
+containerized Claude Code. It never runs the host's Claude executable. Model aliases can be
+set in `.env.devin-bridge`; every configured value must keep the
+`devin-cli-agentic/` prefix.
+
+## Validation commands
+
+The reproducible offline path requires no Devin account and has no runtime Internet:
+
+```bash
+./scripts/devin-bridge/test-unit
+./scripts/devin-bridge/test-contract
+./scripts/devin-bridge/test-e2e-mock
+./scripts/devin-bridge/verify-anthropic-isolation
+```
+
+The authenticated opt-in live path is:
+
+```bash
+ENABLE_LIVE_DEVIN_TESTS=1 ./scripts/devin-bridge/test-live-devin
+```
+
+The live runner waits between scenarios to avoid opening ACP sessions in a burst and
+validates structured Claude stream events instead of trusting textual claims. Its three
+scenarios prove:
+
+1. direct project reads and defect analysis;
+2. a real `Edit`, a client-owned `Bash` `npm test`, and a terminal result;
+3. `/bridge-check` plus `bridge-proof` discovery, project reads, another successful
+   client-owned `npm test`, and completion without pending work.
+
+The final gate also checks the Devin network audit and requires the Claude egress audit to
+remain empty.
+
+## Updating pinned tools
+
+The image pins Node, Claude Code, and Devin CLI in
+`docker/devin-bridge/Dockerfile`. To update:
+
+1. change the explicit versions;
+2. replace both architecture-specific Devin archive checksums with values for the official
+   artifact;
+3. rebuild and run every offline validation command;
+4. confirm the versions inside the image;
+5. rerun the authenticated three-scenario live suite.
+
+Do not install either CLI globally on the host or replace checksum verification with an
+unverified download.
+
+## Diagnosis and cleanup
+
+- `docker compose -f docker/devin-bridge/compose.yml --profile offline logs omniroute`
+  shows local routing and sanitized executor errors.
+- `.sandbox/evidence/mock-acp.jsonl` records deterministic mock ACP actions.
+- `.sandbox/evidence/claude-stream.jsonl` records the real Claude Code offline run.
+- `.sandbox/evidence/live-*.jsonl` records the three validated live streams.
+- `.sandbox/evidence/egress.jsonl` and `.sandbox/evidence/claude-egress.jsonl` are validated,
+  token-free copies of the guard audits.
+
+Stop owned containers and networks while preserving login/config volumes:
+
+```bash
+./scripts/devin-bridge/clean
+```
+
+Remove the complete bridge-owned environment, including named volumes:
+
+```bash
+./scripts/devin-bridge/clean --all
+```
+
+## Limits
+
+- The bridge relies on the fixed no-tools `summarizer` role because Devin CLI `3000.2.17`
+  does not expose a neutral no-tools ACP agent. The adapter compensates for summary-shaped
+  intermediate responses, but one bounded repair can still fail explicitly.
+- Live ACP calls can return transient `502`/`504` responses. The harness spaces scenarios;
+  persistent failure remains fail-closed and never selects another provider.
+- ACP context is reconstructed from each Anthropic request; there is no process/session
+  affinity.
+- One tool call is supported per model response; parallel calls are rejected.
+- Images are explicitly unsupported. Vision, thinking output, effort controls, and a 1M
+  context window are not advertised.
+- SSE uses valid Anthropic lifecycle events but is emitted after the bounded ACP turn is
+  collected; ACP chunks are not forwarded incrementally.

docs/DEVIN_CLAUDE_BRIDGE_PROGRESS.md

@@ -0,0 +1,115 @@
+# Devin Claude Bridge Progress
+
+Updated: 2026-07-28
+
+## Baseline
+
+- Fork version: `3.8.49`.
+- Starting branch: `release/v3.8.49`.
+- Starting commit: `ed7db3ee5f89a144b2d931d8605534522f83de30`.
+- Fixed runtime artifacts: Node `26.0.0`, Claude Code `2.1.220`, Devin CLI `3000.2.17`.
+- Existing `devin-cli` remains unchanged; the new path is the separate
+  `devin-cli-agentic` provider.
+
+## Implemented architecture
+
+- Claude Code runs only inside the non-root bridge container with its own empty config
+  volume and local OmniRoute base URL.
+- `devin-cli-agentic` preserves Anthropic messages, tool schemas, `tool_use`, and
+  `tool_result`, then calls the official Devin CLI over ACP stdio.
+- The executor starts `devin acp --agent-type summarizer`. This is the only fixed official
+  ACP role in the pinned CLI that has no Devin-owned tools.
+- The request is framed as an execution trace. Devin can return one strict client tool
+  envelope; Claude Code executes that tool locally.
+- Internal ACP `tool_call` events, unsupported blocks, invalid schemas, narrative actions,
+  timeouts, cancellation, and process failure all fail closed.
+- Provider and network policy prevent combo/auto/Anthropic fallback.
+
+## Offline proof
+
+- Focused serializer, parser, executor, ACP lifecycle, wire-format, environment, and audit
+  tests pass (39/39).
+- The contract suite covers Anthropic JSON/SSE, `tool_use`, `tool_result` continuation,
+  fragmented ACP frames, stderr, early exit, timeout, cancellation, and fail-closed provider
+  loss.
+- The production bridge image builds with the pinned CLIs.
+- Real Claude Code offline E2E loads `CLAUDE.md`, the project skill and slash command, fires
+  hooks, executes local tools over multiple turns, observes a failed test, repairs the file,
+  reruns the test, and completes.
+- The isolation verifier proves non-root/read-only execution, isolated mounts and config,
+  blocked Anthropic/Claude access, no host credential mounts, local-only inference, and no
+  fallback.
+
+Evidence is generated under `.sandbox/evidence` and ignored by Git.
+
+## Regression status
+
+- `typecheck:core`, focused ESLint, Prettier, shell/Node syntax, and the complete documentation
+  accuracy suite pass.
+- The broad `npm run check` is not reported as passed: after its lint phase, the repository
+  test runner remained alive while an existing `ioredis` client repeatedly retried an
+  unavailable local Redis endpoint after `quota-redis-store.test.ts`. The bridge-focused
+  suites, production image build, offline E2E, isolation proof, and live gate do not use that
+  Redis service and all pass.
+
+## Live Devin proof
+
+Passed with the official in-container login and discovered model
+`swe-1-7-lightning`. The terminal live run completed all three scenarios:
+
+1. Claude Code loaded the fixture instructions, issued client-owned `Read` calls, and
+   returned a correct defect analysis.
+2. Claude Code issued a real `Edit` changing subtraction to addition, then a client-owned
+   `Bash` call running `npm test`; the test reported one pass and zero failures.
+3. Claude Code initialization listed `bridge-check` and `bridge-proof`, read the corrected
+   source and test, executed another client-owned `npm test`, and completed successfully.
+
+The live evidence validator parses stream JSON and requires successful tool results. It does
+not accept a textual claim that a tool ran. It also rejects terminal summaries that report a
+blocker, incomplete work, or required next steps.
+
+The final live gate reported:
+
+```text
+PASS: validated Claude evidence for LIVE_ANALYSIS_COMPLETE
+PASS: validated Claude evidence for LIVE_FIX_COMPLETE
+PASS: validated Claude evidence for BRIDGE_E2E_COMPLETE
+PASS: three live Devin-backed Claude Code scenarios completed
+PASS: live model swe-1-7-lightning was discovered and validated by three scenarios
+```
+
+The same gate validated the network audit: only the Devin guard path was used, no internal
+Devin tool event was accepted, and the Claude egress audit remained empty.
+
+## Investigation conclusion
+
+The initial default-agent hypothesis failed because ACP permission modes do not turn the
+default Devin agent into a raw inference backend. Even `ask` mode can emit Devin-owned
+`tool_call` events. A discovered `allowed-tools: []` agent configuration was not consumed by
+`devin acp` in CLI `3000.2.17`.
+
+The working adaptation uses the official `summarizer` agent because it is structurally
+no-tools. Its fixed summarization behavior can produce intermediate prose, so the bridge
+frames requests as execution traces, detects future-action narration, performs at most one
+strict repair, and otherwise fails. Live validation also exposed transient ACP timeouts;
+the harness now spaces independent scenarios rather than weakening routing or retrying into
+another provider.
+
+## Safety record
+
+No host Claude executable, configuration, login, OAuth token, Keychain, or Anthropic API was
+used. The dedicated Docker volumes remain role-separated. No credential value is written to
+the repository or evidence output.
+
+During the early baseline, a focused test without isolated `DATA_DIR` initialized the
+repository's normal OmniRoute database at `/Users/lucasisrael/.omniroute/storage.sqlite`.
+It was not rolled back or touched again. Every bridge command now pins database and temporary
+paths under the worktree's `.sandbox` directory.
+
+## Remaining limits
+
+- The no-tools backend has a summarizer system role rather than a neutral generation role.
+- One client tool call per response is supported; parallel tool calls are rejected.
+- ACP processes are per-turn and stateless.
+- Live Devin availability can still produce explicit `502`/`504` failures.
+- Images and unadvertised vision/effort/large-context capabilities remain unsupported.

docs/INCIDENT_RESPONSE.md

@@ -1,3 +1,9 @@
+---
+title: "Incident Response Runbook — OmniRoute (2026-06-18)"
+version: 3.8.50
+lastUpdated: 2026-08-06
+---
+
 # Incident Response Runbook — OmniRoute (2026-06-18)
 
 **Status**: Authoritative. The 71-pillar audit (L61) references this doc

docs/PERF_BUDGETS.md

@@ -1,3 +1,9 @@
+---
+title: "Performance Budgets — OmniRoute (2026-06-18)"
+version: 3.8.50
+lastUpdated: 2026-08-06
+---
+
 # Performance Budgets — OmniRoute (2026-06-18)
 
 **Status**: Authoritative. SLO targets that the 71-pillar audit (L13)

docs/README.md

@@ -105,6 +105,7 @@ Pluggable subsystems exposed to clients, agents, and operators.
 - [PLUGINS.md](frameworks/PLUGINS.md) — CLI plugin system overview.
 - [PLUGIN_SDK.md](frameworks/PLUGIN_SDK.md) — plugin SDK reference.
 - [PLUGIN_MARKETPLACE.md](frameworks/PLUGIN_MARKETPLACE.md) — plugin marketplace.
+- [RADAR.md](frameworks/RADAR.md) — Radar free-model catalog overlay (optional, off by default).
 
 ## routing/
 

docs/ROADMAP.md

@@ -1,3 +1,9 @@
+---
+title: "OmniRoute Roadmap"
+version: 3.8.50
+lastUpdated: 2026-08-06
+---
+
 # OmniRoute Roadmap
 
 > Version-gated, not date-gated: each milestone ships when its quality gates pass.

docs/architecture/REPOSITORY_MAP.md

@@ -184,6 +184,7 @@ src/
 | `config/`                                | Runtime config helpers                                                                                                                                                                                                                                                                                                  |
 | `db/`                                    | 95+ domain DB modules + 110+ migrations (always go through here for SQLite)                                                                                                                                                                                                                                             |
 | `quota/`                                 | Quota Sharing Engine: `dimensions.ts` (types/Zod), `types.ts` (QuotaStore interface), `sqliteQuotaStore.ts`, `redisQuotaStore.ts`, `storeFactory.ts`, `fairShare.ts`, `burnRate.ts`, `planResolver.ts`, `planRegistry.ts`, `saturationSignals.ts`, `enforce.ts`, `spendRecorder.ts` — see `docs/routing/QUOTA_SHARE.md` |
+| `radar/`                                 | Radar free-model catalog client: `feedSchema.ts`, `pinnedKeys.ts`, `verify.ts`, `sync.ts`, `applyFeed.ts`, `index.ts` (`getRadarCatalog()`) — see `docs/frameworks/RADAR.md`                                                                                                                                            |
 | `display/`                               | UI formatting helpers (cost, latency, etc.)                                                                                                                                                                                                                                                                             |
 | `embeddings/`                            | Embeddings service helpers                                                                                                                                                                                                                                                                                              |
 | `env/`                                   | Env variable parsing + validation                                                                                                                                                                                                                                                                                       |
@@ -410,6 +411,7 @@ open-sse/
 | `AGENT_PROTOCOLS_GUIDE.md` | Unified guide: A2A vs ACP vs Cloud Agents                           |
 | `CLOUD_AGENT.md`           | Codex Cloud / Devin / Jules orchestration                           |
 | `SKILLS.md`                | Skills framework (built-in + marketplace + SkillsSH + sandbox)      |
+| `RADAR.md`                 | Radar free-model catalog overlay (`RADAR_ENABLED`, off by default)  |
 | `MEMORY.md`                | Memory system (SQLite FTS5 + Qdrant)                                |
 | `EVALS.md`                 | Eval framework (suites, runs, rubrics)                              |
 | `GUARDRAILS.md`            | PII masker, prompt injection, vision bridge                         |

docs/assets/flags/README.md

@@ -1,3 +1,9 @@
+---
+title: "Flag icons"
+version: 3.8.50
+lastUpdated: 2026-08-06
+---
+
 # Flag icons
 
 SVG country flags used by the language selector in the root `README.md`.

docs/combo-context-requirements.md

@@ -1,3 +1,9 @@
+---
+title: "Combo Context Requirements Feature"
+version: 3.8.50
+lastUpdated: 2026-08-06
+---
+
 # Combo Context Requirements Feature
 
 ## Overview

docs/compression/COMPRESSION_LANGUAGE_PACKS.md

@@ -7,7 +7,7 @@ lastUpdated: 2026-06-28
 # Compression Language Packs
 
 Caveman compression can load language-specific rule packs in addition to the built-in English rules.
-This keeps the core engine stable while allowing Portuguese, Spanish, German, French, Japanese, and
+This keeps the core engine stable while allowing Portuguese, Spanish, German, French, Italian, Japanese, and
 future language packs to evolve independently.
 
 ## Location
@@ -26,11 +26,12 @@ Current shipped packs (verified against `rules/` directory contents):
 | Spanish             | `rules/es/`    | `context`, `dedup`, `filler`, `structural`, `ultra` |
 | Portuguese (Brazil) | `rules/pt-BR/` | `context`, `dedup`, `filler`, `structural`, `ultra` |
 | Indonesian          | `rules/id/`    | `context`, `dedup`, `filler`, `structural`, `ultra` |
+| Italian             | `rules/it/`    | `context`, `dedup`, `filler`, `structural`, `ultra` |
 | German              | `rules/de/`    | `context`, `filler`, `structural`                   |
 | French              | `rules/fr/`    | `context`, `filler`, `structural`                   |
 | Japanese            | `rules/ja/`    | `context`, `filler`, `structural`                   |
 
-> **Parity note:** `en`, `es`, `pt-BR`, and `id` packs have the full 5 categories; `de`, `fr`, `ja` ship 3 categories. The missing `dedup` and `ultra` categories silently fall back to the English built-ins. Contributions welcome to add `dedup.json` and `ultra.json` for the smaller packs.
+> **Parity note:** `en`, `es`, `pt-BR`, `id`, and `it` packs have the full 5 categories; `de`, `fr`, `ja` ship 3 categories. The missing `dedup` and `ultra` categories silently fall back to the English built-ins. Contributions welcome to add `dedup.json` and `ultra.json` for the smaller packs.
 >
 > The `pt-BR` pack is based on **[Troglodita](https://github.com/leninejunior/troglodita)** by Lenine Júnior — a compression system designed from scratch for Brazilian Portuguese grammar (pleonasm reduction, PT-BR filler removal, technical abbreviations for the dev BR community).
 >

docs/frameworks/MCP-SERVER.md

@@ -6,7 +6,7 @@ lastUpdated: 2026-06-28
 
 # OmniRoute MCP Server Documentation
 
-> Model Context Protocol server with 104 tools across routing, cache, compression, memory, skills, proxy, pool, and context source operations.
+> Model Context Protocol server with 105 tools across routing, cache, compression, memory, skills, proxy, pool, and context source operations.
 >
 > Source of truth: `open-sse/mcp-server/server.ts` computes **104 unique tools** with `countUniqueMcpTools()`: 42 canonical definitions (including the six CCR lifecycle tools and the agent-skills trio), plus memory (3), skills (4), GitHub skills (3), pool (6), gamification (8), plugins (8), Notion (6), Obsidian (22), and two RTK-only compression tools.
 
@@ -369,7 +369,7 @@ MCP tool, prompt, and resource registries can compress descriptions at registrat
 
 Description compression shrinks each tool's metadata; **tool-cardinality reduction** goes one step further by reducing _how many_ tools are announced at all. Advertising fewer tools in the `tools/list` manifest cuts the per-request token cost the client's model pays for the tool catalog ("layer 5" compression). The implementation is a pure, stateless filter in `open-sse/mcp-server/toolCardinality.ts` (`reduceToolManifest`), wired into the registration loop in `createMcpServer()` (`open-sse/mcp-server/server.ts`).
 
-**Opt-in, off by default.** The filter only runs when at least one of two environment variables is set; with neither set, all 104 tools are announced unchanged.
+**Opt-in, off by default.** The filter only runs when at least one of two environment variables is set; with neither set, all 105 tools are announced unchanged.
 
 | Variable         | Mode                                                                                    |
 | :--------------- | :-------------------------------------------------------------------------------------- |

docs/frameworks/MEMORY_BACKEND.md

@@ -0,0 +1,228 @@
+---
+title: "MemoryBackend Provider Pattern"
+version: 3.8.49
+lastUpdated: 2026-07-28
+---
+
+# MemoryBackend Provider Pattern
+
+> **Source of truth:** `src/lib/memory/backend.ts`, `src/lib/memory/genericBackend.ts`, `src/lib/memory/manager.ts`
+> **Tests:** `src/lib/memory/__tests__/generic-backend.test.ts`
+
+The MemoryBackend provider pattern introduces a **pluggable backend abstraction layer** over the existing memory engine. Instead of being tied to a single storage implementation, the memory system now supports multiple backends (SQLite, Obsidian, Notion, custom HTTP backends) with configurable primary/fallback routing.
+
+## Architecture
+
+```
+┌──────────────────────────────────────────────────────────┐
+│                    API Routes                             │
+│            (src/app/api/memory/route.ts)                  │
+└──────────────────────┬───────────────────────────────────┘
+                       │
+┌──────────────────────▼───────────────────────────────────┐
+│                   MemoryManager                           │
+│           Singleton orchestrator (manager.ts)             │
+│                                                          │
+│  Primary ──► Backend A  (e.g. SQLite)                    │
+│  Fallback ─► Backend B  (e.g. Obsidian)                  │
+│             Backend C  (e.g. Notion via GenericBackend)   │
+└──────────────────────┬───────────────────────────────────┘
+                       │
+        ┌──────────────┼──────────────┐
+        ▼              ▼              ▼
+┌────────────┐ ┌────────────┐ ┌──────────────────┐
+│ SQLite     │ │ Obsidian   │ │ GenericMemory    │
+│ Backend    │ │ Backend    │ │ Backend (HTTP)   │
+└────────────┘ └────────────┘ └──────────────────┘
+```
+
+### Core Interface (`backend.ts`)
+
+Every backend must implement the `MemoryBackend` interface:
+
+```typescript
+interface MemoryBackend {
+  readonly id: string;
+  readonly displayName: string;
+
+  // CRUD
+  create(input: CreateMemoryInput): Promise<Memory>;
+  get(id: string): Promise<Memory | null>;
+  update(id: string, updates: Partial<...>): Promise<boolean>;
+  delete(id: string): Promise<boolean>;
+  list(filter: MemoryFilter): Promise<{ data: Memory[]; total: number; byType: Record<string, number> }>;
+
+  // Search
+  search(config: SearchConfig): Promise<Memory[]>;
+
+  // Health
+  health(): Promise<HealthCheckResult>;
+
+  // Lifecycle (optional)
+  initialize?(): Promise<void>;
+  shutdown?(): Promise<void>;
+}
+```
+
+### MemoryManager (`manager.ts`)
+
+Singleton orchestrator that:
+
+- **Registers** backends via `register(backend)` — called at boot from `index.ts`
+- **Configures** primary + fallback via `configure(primary, fallbacks)`
+- **Routes** CRUD/search to the primary, with fallback chain on failure
+- **Health checks** all backends periodically
+
+**Fallback behavior:**
+
+| Operation | Primary              | Fallbacks               |
+| --------- | -------------------- | ----------------------- |
+| `create`  | ✅ Primary only      | ❌                      |
+| `get`     | ✅ Try primary first | ✅ Fallback if null     |
+| `update`  | ✅ Primary only      | ✅ Fire-and-forget sync |
+| `delete`  | ✅ Primary only      | ✅ Fire-and-forget sync |
+| `list`    | ✅ Primary only      | ❌                      |
+| `search`  | ✅ Primary first     | ✅ Fallback on error    |
+
+### GenericMemoryBackend (`genericBackend.ts`)
+
+A generic HTTP connector that adapts any REST API into a MemoryBackend. Useful for:
+
+- **Notion** — connect via Notion API
+- **Obsidian** — connect via Obsidian Local REST API
+- **Custom backends** — any service that exposes a RESTful memory API
+
+**Configuration:**
+
+```typescript
+interface GenericBackendConfig {
+  baseUrl: string;           // Base URL of the backend API
+  apiKey?: string;           // Bearer token for auth
+  headers?: Record<string, string>;  // Custom HTTP headers
+  timeout?: number;          // Request timeout (default: 30000ms)
+  backendType?: string;      // For logging
+
+  // Endpoint overrides (defaults use REST conventions)
+  endpoints?: {
+    search?: string;   // default: "/memories/search"
+    create?: string;   // default: "/memories"
+    list?: string;     // default: "/memories"
+    get?: string;      // default: "/memories/{id}"
+    update?: string;   // default: "/memories/{id}"
+    delete?: string;   // default: "/memories/{id}"
+    health?: string;   // default: "/health"
+  };
+
+  // Query parameter name mappings
+  queryParams?: {
+    query?/apiKeyId?/limit?/offset?/strategy?/maxTokens?/type?/sessionId?/orderBy?/orderDir?/options?
+  };
+
+  // Path parameter name mappings
+  pathParams?: {
+    id?/memoryId?
+  };
+}
+```
+
+**Known backends** are pre-configured in `KNOWN_BACKENDS`:
+
+```typescript
+createKnownBackend("obsidian"); // → GenericMemoryBackend pointed at localhost:27123
+createKnownBackend("notion"); // → GenericMemoryBackend pointed at api.notion.com/v1
+```
+
+### Built-in Backends
+
+#### SQLiteBackend (`sqliteBackend.ts`)
+
+The default primary backend. Wraps the existing SQLite-based memory store using `src/lib/memory/store.ts`. Automatically registered at boot.
+
+```typescript
+import { sqliteBackend } from "./sqliteBackend";
+memoryManager.register(sqliteBackend);
+```
+
+#### ObsidianBackend (`obsidianBackend.ts`)
+
+Wraps the existing Obsidian integration (`src/lib/memory/obsidianBackend.ts`). Connects to an Obsidian vault via the Obsidian Local REST API.
+
+## Settings
+
+Memory backend settings are stored in the app settings table and managed via `src/lib/memory/settings.ts`:
+
+| Setting           | Env/Config Key           | Default    | Description                  |
+| ----------------- | ------------------------ | ---------- | ---------------------------- |
+| Primary backend   | `memoryPrimaryBackend`   | `"sqlite"` | ID of the primary backend    |
+| Fallback backends | `memoryFallbackBackends` | `[]`       | Ordered fallback backend IDs |
+| Backend configs   | `memoryBackendConfigs`   | `{}`       | Per-backend config overrides |
+
+Settings are normalized via `normalizeMemorySettings()` and cached at `getMemorySettings()`.
+
+## Initialization Flow
+
+```
+App bootstrap
+  → index.ts imports (side-effect): registers SQLiteBackend
+  → initMemoryBackends() called from app lifecycle:
+      1. Load settings (getMemorySettings)
+      2. Configure primary + fallback
+      3. Initialize all backends (health check)
+      4. Ready for requests
+```
+
+## Adding a New Backend
+
+1. **Implement `MemoryBackend`** interface in `src/lib/memory/<name>Backend.ts`
+2. **Export** from `src/lib/memory/index.ts`
+3. **Register** with `memoryManager.register(yourBackend)` at boot
+4. **Configure** via settings: set `memoryPrimaryBackend` to your backend ID
+5. **Test** with `src/lib/memory/__tests__/generic-backend.test.ts` as reference
+
+### Example: Brain Backend
+
+```typescript
+import { createGenericMemoryBackend } from "./genericBackend";
+
+const brainBackend = createGenericMemoryBackend("brain", "BK-Brain", {
+  baseUrl: process.env.BRAIN_API_URL || "http://localhost:9099",
+  apiKey: process.env.BRAIN_API_KEY,
+  endpoints: {
+    search: "/api/memory/search",
+    create: "/api/memory",
+    health: "/api/health",
+  },
+});
+
+memoryManager.register(brainBackend);
+```
+
+## Verification
+
+### Unit tests
+
+```bash
+npx vitest run src/lib/memory/__tests__/generic-backend.test.ts --reporter=verbose
+```
+
+Expected output: **26 tests, all passing** covering:
+
+- Constructor (2)
+- Health check (4) — success, failure 500, network error, latency
+- Initialize (2) — success, failure
+- Create (2) — default endpoint, custom endpoint
+- Get (4) — success, 404 → null, non-404 throw, custom path params
+- Update (2) — success, 404 → false
+- Delete (2) — success, 404 → false
+- List (2) — query params, custom param names
+- Search (3) — query params, custom endpoint, options serialization
+- Auth headers (2) — Bearer token, custom headers
+- Factory (1)
+
+### Type check
+
+```bash
+npm run typecheck:core
+```
+
+Expected: **0 errors**.

docs/frameworks/RADAR.md

@@ -0,0 +1,385 @@
+---
+title: "Radar Free-Model Catalog"
+version: 3.8.50
+lastUpdated: 2026-08-07
+---
+
+# Radar Free-Model Catalog
+
+> **Source of truth:** `src/lib/radar/`, `src/lib/db/radar.ts`, `src/app/api/radar/`
+> **Last updated:** 2026-08-07 — v3.8.50
+
+Radar is an **optional add-on** that overlays a signed, freshly-curated free-model
+catalog on top of the release baseline (`FREE_MODEL_BUDGETS` in
+`open-sse/config/freeModelCatalog.data.ts`). It exists because the free-tier landscape moves
+faster than release cadence — providers add, shrink, or discontinue free quotas between
+releases, and the baseline catalog can only be refreshed when a new version ships.
+
+**Nothing that is free today stops being free.** Radar never removes or paywalls a
+baseline entry; it only refreshes limits/status fields at read time and can layer in
+newly-discovered free models between releases. The baseline catalog itself is never
+mutated on disk — see [Read-time overlay merge rules](#read-time-overlay-merge-rules)
+below.
+
+---
+
+## Flag: `RADAR_ENABLED` (default off)
+
+Radar is gated end-to-end by the `RADAR_ENABLED` feature flag
+(`src/shared/constants/featureFlagDefinitions.ts`, category `policies`,
+`defaultValue: "false"`).
+
+**When the flag is off, the surface does not exist:**
+
+- `GET /api/radar/catalog`, `POST /api/radar/sync`, `POST /api/radar/settings` all
+  return `404` before touching any Radar module.
+- The dashboard screens (`/dashboard/radar`, `/dashboard/radar/setup`) render
+  `notFound()`.
+- `getRadarCatalog()` (`src/lib/radar/index.ts`) returns the untouched baseline —
+  same entry count, same values, every entry tagged `origin: "baseline"` — and never
+  reads the feed cache.
+- No network call is ever made; `syncRadar()` (`src/lib/radar/sync.ts`) returns
+  `{ status: "disabled" }` at step 1 without touching `fetch`.
+
+This is a strict superset gate: flipping the flag on unlocks the _screens_, nothing
+more. It does not upload data, does not start a background sync, and does not change
+routing or model selection — see the separate opt-in below.
+
+---
+
+## Data sync is a SEPARATE opt-in — the privacy promise
+
+Turning `RADAR_ENABLED` on only unlocks the UI. Syncing the feed requires a second,
+independent opt-in stored in `radar_settings.opt_in` (`src/lib/db/radar.ts`,
+migration `136_radar_cache_settings.sql`). `syncRadar()` checks the flag _and_ the
+opt-in before making any network call:
+
+```
+Flag off      → { status: "disabled" }   — no network call
+Opt-in false  → { status: "opt_out" }    — no network call
+```
+
+When both are on, the sync path is:
+
+1. `GET <feed base URL>/v1/catalog/latest` with an optional `Authorization: Bearer
+<supporter key>` header (see below).
+2. Nothing about the request, the operator, or their traffic is uploaded — it is a
+   plain, unauthenticated-by-default GET. OmniRoute never posts usage data, provider
+   configuration, or model traffic to the feed service.
+3. The response is verified, validated, and cached locally (see
+   [Security model](#security-model)). Nothing else touches the network for Radar.
+
+The **supporter key** is an optional Bearer token (`radar_settings.supporter_key`)
+that lets the feed service decide which tier to serve (see
+[Tiers](#tiers-community-and-live)). It is:
+
+- Stored **encrypted at rest** with the same AES-256-GCM `encrypt()`/`decrypt()`
+  helpers (`src/lib/db/encryption.ts`) used for provider credentials.
+- Set via `POST /api/radar/settings` (`{ supporterKey: "omr_" + 40 hex chars }`) and
+  **never echoed back** — the response returns a masked form (`omr_****abcd`).
+- Sent to the feed service as a Bearer token on the sync GET — nothing else about the
+  key ever leaves the client.
+
+---
+
+## Security model
+
+### Ed25519 signature over exact bytes
+
+The feed payload is signed with Ed25519. `verifyFeedBytes()`
+(`src/lib/radar/verify.ts`) verifies the signature over the **exact response bytes**
+received over the wire — the payload is never re-serialized before verification, so a
+byte-for-byte re-encoding cannot silently invalidate or bypass the signature check.
+Verification failure (`invalid_signature`) aborts the sync before the payload is ever
+parsed or cached.
+
+### Pinned public key + rotation
+
+The verifying public key is pinned in `src/lib/radar/pinnedKeys.ts`
+(`PINNED_FEED_PUBLIC_KEYS`), an array so a new key can be prepended ahead of a
+rotation while old cached feeds signed with a previous key remain valid until
+re-synced.
+
+### Fork-friendly env overrides
+
+Two env vars let forks and self-hosters point the client at their own feed instead of
+the default OmniRoute service — see
+[How to self-host a feed](#how-to-self-host-a-feed) below:
+
+| Var                 | Purpose                                                                                                      |
+| ------------------- | ------------------------------------------------------------------------------------------------------------ |
+| `RADAR_FEED_URL`    | Overrides the feed base URL (default `https://radar.omniroute.online`).                                      |
+| `RADAR_FEED_PUBKEY` | Overrides the pinned public key (base64-DER SPKI or PEM), replacing the built-in array with this single key. |
+
+### Version floor
+
+`syncRadar()` rejects a downloaded feed whose `version` is not strictly newer than the
+currently cached version (`compareVersions()`, dotted `YYYY.MM.DD.n` comparison) —
+`{ status: "stale" }`. This prevents a compromised or misconfigured feed endpoint from
+rolling a client back to an older, differently-signed payload.
+
+### Schema validation
+
+The downloaded bytes are parsed and validated against `RadarFeedSchema`
+(`src/lib/radar/feedSchema.ts`, a Zod schema) **after** signature verification. A
+schema mismatch returns `{ status: "invalid_schema" }` and the cache is left
+untouched. The cached payload is defensively re-validated again on every read
+(`getRadarCatalog()`) — a corrupted or hand-edited cache row falls back to the
+baseline rather than being served.
+
+### Response size cap (10 MB)
+
+`syncRadar()` enforces a **10 MB hard cap** on the feed response body — the signed
+feed is a KB-scale JSON document, so anything past this points at a misconfigured or
+hostile `RADAR_FEED_URL` (or an upstream serving garbage), not a legitimate catalog.
+Enforcement is two-layered:
+
+1. A `Content-Length` preflight check skips reading the body entirely when the
+   header already declares a value over the cap.
+2. A running-total check while reading the body enforces the cap even when
+   `Content-Length` is absent or understates the real size — the header is never
+   trusted on its own. Concatenating the accumulated chunks preserves the exact
+   bytes needed for the Ed25519 signature check afterward.
+
+Exceeding the cap returns `{ status: "too_large" }` and leaves the cache untouched,
+following the same non-destructive pattern as every other sync failure
+(`invalid_signature`, `invalid_schema`, `stale`).
+
+---
+
+## Tiers: `community` and `live`
+
+The feed schema carries a `tier: "community" | "live"` field, decided **server-side**
+by the feed service based on the request (presence and validity of the supporter key)
+— the client never decides its own tier.
+
+- **`community`** — the free catalog delayed by roughly 30 days behind the freshest
+  data. This is what an unauthenticated or invalid-key request receives.
+- **`live`** — the freshest catalog, served to requests carrying a valid supporter
+  key.
+
+**An invalid or expired supporter key degrades to `community` — it is never an
+error.** The sync path only distinguishes signature/schema/version failures (all
+recoverable, all non-fatal to the cached state) from a successful `{ status:
+"updated", version, tier }`. There is no tier-specific error path a client needs to
+handle.
+
+### The served tier comes from a response header, not the signed body
+
+The signed feed **body**'s `tier` field is always `"live"` — the feed service ships
+**one signed artifact per version**, so the body cannot carry a per-request tier
+without invalidating the Ed25519 signature (re-signing per request would defeat the
+point of a pinned, cacheable, verifiable artifact). The tier actually served for a
+given request is instead carried in the **`x-omniroute-feed-tier` response header**,
+decided server-side from the request's `Authorization` key.
+
+`syncRadar()` (`src/lib/radar/sync.ts::parseServedTierHeader()`) is the single place
+that resolves the tier a client should trust:
+
+1. Parse `x-omniroute-feed-tier` with `RadarTierSchema` (Zod) — an absent header, or
+   a value that isn't exactly `"community"` or `"live"`, is treated as **not
+   present** (never trusted into the cache/UI as-is; this also covers older feed
+   servers that predate the header).
+2. Fall back to the signed body's `tier` field (always `"live"`) only when step 1
+   yields nothing.
+3. The resolved tier is what gets cached and returned as `{ status: "updated",
+   version, tier }` — this is the value the dashboard shows, never the raw body
+   field.
+
+---
+
+## Read-time overlay merge rules
+
+`applyFeed()` (`src/lib/radar/applyFeed.ts`) merges the cached feed **over** the
+static baseline at **read time**, inside `getRadarCatalog()`. The baseline array
+(`FREE_MODEL_BUDGETS`) is never mutated — a `MergedEntry[]` is computed fresh on every
+call.
+
+Four rules, in order of precedence:
+
+1. **Feed never overwrites a local override.** Per-field: if the operator has
+   customized a field on an entry (`localOverrides` map, keyed `provider:modelId`),
+   the feed's value for that specific field is skipped — the operator's value wins.
+2. **`enabled: false` disables the entry, with provenance.** A feed entry that turns
+   an entry off sets `enabled: false` and `disabledBy: "radar"` on the merged result,
+   so the UI can explain _why_ an entry went from available to disabled.
+3. **A user-added entry not present in the feed survives untouched.** Entries that
+   only exist in the baseline (or were added locally) and have no corresponding feed
+   entry pass through unchanged.
+4. **A tombstoned entry is never resurrected.** If the operator explicitly deleted an
+   entry (`tombstones` set), the feed re-adding that `provider:modelId` in a later
+   version does not bring it back.
+
+### Provenance markers
+
+Every merged entry carries an `origin` field the UI renders as a badge:
+
+- `"baseline"` — untouched from the static release catalog.
+- `"radar"` — one or more fields were refreshed by the feed.
+- `"local"` — the operator has at least one local override on this entry (local
+  overrides always win over the feed per rule 1, regardless of what the feed says).
+
+---
+
+## Local surfaces — never a feed proxy
+
+Five local routes back the UI, all under `src/app/api/radar/`:
+
+| Route                   | Method | Purpose                                                                                          |
+| ----------------------- | ------ | -------------------------------------------------------------------------------------------------- |
+| `/api/radar/catalog`    | GET    | Returns the merged catalog (`getRadarCatalog()`) from the local cache.                            |
+| `/api/radar/sync`       | POST   | Triggers `syncRadar()` server-side; returns the resulting status.                                 |
+| `/api/radar/settings`   | GET    | Returns `{ optIn, hasSupporterKey, supporterKeyMasked }` — never the raw key.                     |
+| `/api/radar/settings`   | POST   | Sets opt-in and/or the (encrypted) supporter key.                                                 |
+| `/api/radar/referrals`  | GET    | Returns `{ fixed, campaigns, tier }` from the local cache — see [Referral links](#referral-links-free-credits) below. |
+
+**Hard rule: these routes never proxy the feed service.** The browser only ever talks
+to the local OmniRoute server; `syncRadar()` is the single module in the whole client
+that touches the network for Radar (`src/lib/radar/sync.ts`), and it always runs
+server-side, never client-side. This keeps the feed URL and any supporter key
+out of client-facing network traffic entirely.
+
+All five routes return `404` when `RADAR_ENABLED` is off (see
+[Flag](#flag-radar_enabled-default-off) above), and route error responses through
+`buildErrorBody()`/`sanitizeErrorMessage()` per the repo-wide error-sanitization rule
+(`docs/security/ERROR_SANITIZATION.md`).
+
+### Authentication
+
+All five routes require authentication via `isAuthenticated()`
+(`src/shared/utils/apiAuth.ts`) — a dashboard session cookie or a management-scoped
+API key, the same gate that protects the rest of `/api/settings/*`. The flag-off
+`404` check always runs **before** the auth check, so an install with `RADAR_ENABLED`
+off stays byte-identical (no auth prompt just to learn the surface doesn't exist);
+once the flag is on, an unauthenticated request gets `401` before any DB read or
+write. `GET /api/radar/settings` never returns the raw supporter key regardless of
+auth state — only the masked form and a `hasSupporterKey` boolean.
+
+---
+
+## Referral links (free credits)
+
+The server-published feed carries a `referrals` section (server-side D28 work, already
+in production — this section documents the **client** consumption only):
+
+```ts
+referrals: {
+  fixed: RadarReferral[],      // present in EVERY tier, including community
+  campaigns: RadarReferral[],  // only populated on the live (supporter) tier;
+                                // the community artifact always publishes []
+}
+// RadarReferral = { provider, url, kind: "fixo" | "campanha", validUntil,
+//                    requiredAction, isDefault }
+```
+
+The client never decides which tier it received or which referrals belong in which
+tier — the server already publishes two artifacts (`live`/`community`) with
+`campaigns` gated server-side, same principle as the [tiers](#tiers-community-and-live)
+section above. `RadarFeedSchema` (`src/lib/radar/feedSchema.ts`) validates `referrals`
+as a whole-object `.default({fixed:[],campaigns:[]})`, and `campaigns` defaults
+independently inside it — so a feed cached before this section existed on the server
+still parses cleanly, and `campaigns` alone can also be absent without failing
+validation. Every `RadarReferral.url` must be `https://` — a `http://` url fails
+schema validation.
+
+### Accessor
+
+`src/lib/radar/index.ts` exports two read-only accessors, both never throwing (same
+defensive contract as `getRadarCatalog()` — flag off, no cache, or a corrupt/old cached
+payload all resolve to the empty shape instead of an error):
+
+- `getRadarReferrals()` → `{ fixed: RadarReferral[], campaigns: RadarReferral[] }`.
+- `getDefaultReferralFor(provider)` → the `fixed` referral with `isDefault: true` for
+  that provider, or `null`. Only looks at `fixed` — a campaign is never used as a
+  provider's "default" link.
+
+The actual "which referral is the default for a provider" rule lives in
+`findDefaultReferral()` (`src/lib/radar/referrals.ts`), a small pure function with **no
+DB import** — it is safe to import into a `"use client"` component. `getRadarReferrals`/
+`getDefaultReferralFor` (in `index.ts`) pull in `@/lib/db/radar` and therefore stay
+server-only; the providers dashboard imports `referrals.ts` directly instead of
+`index.ts` (see below) to avoid bundling `better-sqlite3` into the browser.
+
+### `GET /api/radar/referrals`
+
+Follows the exact same gate order as every other Radar route: `RADAR_ENABLED` off →
+`404` (checked first, byte-identical inertia); unauthenticated → `401`; otherwise `200`
+with `{ fixed, campaigns, tier }` — `tier` comes straight from the cache row and is
+purely informative (drives the UI's soft upsell copy below), the route does no
+gating of its own. Never proxies the feed server — same local-cache-only contract as
+`/api/radar/catalog`.
+
+### Dashboard UI — "Free credits" tab on `/dashboard/radar`
+
+Reuses the existing Radar page (`src/app/(dashboard)/dashboard/radar/page.tsx`) as a
+second tab instead of a new route — less routing/i18n surface for a feature that is a
+variation on data the page already fetches. Once opted in, the tab bar offers
+**Catalog** (existing table) and **Free credits**:
+
+- Fixed links are grouped by provider, each showing `requiredAction` (when present)
+  and a `target="_blank" rel="noopener noreferrer"` button to the referral URL.
+- Campaigns show the same, plus `validUntil` when present.
+- When `campaigns` is empty **and** the served tier is `community`, the UI shows a
+  short upsell note ("limited-time campaigns are a supporter extra") — this **never**
+  hides or gates the fixed links list, which stays fully populated for every tier. The
+  upsell is soft messaging only, never a block.
+
+### Referral link on the provider name (providers dashboard)
+
+`ProviderPageHeader` (`src/app/(dashboard)/dashboard/providers/[id]/components/`)
+already linked the provider name to `providerInfo.website` when present, with one
+precedent for a monetized link: the Kimi (Moonshot AI) partner-link note
+(`providers.kimiPartnerLinkNote` i18n key). D28 reuses that exact same discreet-note
+pattern for Radar default referrals instead of introducing a new key.
+
+Loose coupling, by design:
+
+- `resolveProviderHeaderLink()` (`src/app/(dashboard)/dashboard/providers/providerPageUtils.ts`)
+  is a **pure** function — `(staticWebsite, referralUrl) => { website, isReferralLink }`
+  — with no dependency on `@/lib/radar` or `@/lib/db/*`. `providerPageUtils.ts` as a
+  whole stays free of those imports (asserted by
+  `tests/unit/provider-header-referral-link.test.ts`).
+- `ProviderDetailPageClient.tsx` (a `"use client"` component) is the one place allowed
+  to fetch Radar data — via `fetch("/api/radar/referrals")`, the same local-route
+  pattern the Radar dashboard page itself uses — and it computes the default referral
+  client-side with `findDefaultReferral()` from the DB-free `src/lib/radar/referrals.ts`.
+- With `RADAR_ENABLED` off, the fetch 404s, `referralUrl` stays `null`, and
+  `resolveProviderHeaderLink()` returns the static catalog `website` unchanged — the
+  provider page is byte-identical to before this feature existed. Same outcome when
+  there is no cache yet or no default referral for that specific provider.
+- When a default referral does apply, `ProviderPageHeader` receives `isReferralLink`
+  and shows the same discreet note/tooltip as the Kimi partner link (reusing the
+  `providers.kimiPartnerLinkNote` key) — never a new, separate visual treatment.
+
+---
+
+## How to self-host a feed
+
+A fork or self-hoster that wants full control over the catalog can run their own feed
+service without touching client code:
+
+1. Serve a `GET /v1/catalog/latest` endpoint returning a JSON body that satisfies
+   `RadarFeedSchema` (`src/lib/radar/feedSchema.ts`) — top-level `feed:
+"omniroute-radar"`, `schemaVersion: 1`, `version`, `tier`, `providers`, `models`,
+   `quirks`, and `totals`.
+2. Sign the exact response bytes with an Ed25519 key pair and return the base64
+   signature in the `x-omniroute-feed-signature` response header.
+3. Set `RADAR_FEED_URL` to the new base URL and `RADAR_FEED_PUBKEY` to the matching
+   public key (base64-DER SPKI or PEM) — see the
+   [env var reference](../reference/ENVIRONMENT.md#27-radar-feed-self-hosting).
+4. Enable `RADAR_ENABLED` and opt in via `POST /api/radar/settings`
+   (`{ optIn: true }`).
+
+No other code changes are required — `verifyFeedBytes()` picks up the override
+automatically (`getFeedPublicKeys()` in `src/lib/radar/pinnedKeys.ts`), and version
+comparison, schema validation, and the merge rules apply identically to a self-hosted
+feed.
+
+---
+
+## Related docs
+
+- [`docs/security/ERROR_SANITIZATION.md`](../security/ERROR_SANITIZATION.md) — the
+  error-response pattern the five `/api/radar/*` routes follow.
+- [`docs/reference/ENVIRONMENT.md`](../reference/ENVIRONMENT.md#27-radar-feed-self-hosting)
+  — `RADAR_FEED_URL` / `RADAR_FEED_PUBKEY` reference.

docs/getting-started/AUTO-COMBO-GUIDE.md

@@ -1,3 +1,9 @@
+---
+title: "Auto-Combo: Let OmniRoute Pick the Best AI for You"
+version: 3.8.50
+lastUpdated: 2026-08-06
+---
+
 # Auto-Combo: Let OmniRoute Pick the Best AI for You
 
 > **TL;DR**: Set your model to `auto` and OmniRoute automatically picks the best AI provider for each request. No configuration needed.