build(deps): bump github/codeql-action from 4.37.8 to 4.37.9

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 4.37.8 to 4.37.9. - [Release notes](https://github.com/github/codeql-action/releases) - [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md) - [Commits](https://github.com/github/codeql-action/compare/v4.37.8...v4.37.9) --- updated-dependencies: - dependency-name: github/codeql-action dependency-version: 4.37.9 dependency-type: direct:production update-type: version-update:semver-patch ... Signed-off-by: dependabot[bot] <support@github.com>
chore(deps): freeze onnxruntime-node and eslint-plugin-react-hooks out of dependabot groups (#12329 )
2026-09-02 04:42:17 +03:00 · 2026-09-01 18:59:09 +00:00 · 2026-09-01 15:54:04 -03:00 · 2026-09-01 15:37:48 -03:00 · 2026-09-01 15:36:39 -03:00 · 2026-09-01 12:08:31 -03:00
230 changed files with 5906 additions and 2822 deletions
--- a/.env.example
+++ b/.env.example
@@ -1470,25 +1470,17 @@ CURSOR_USER_AGENT="Cursor/3.4"
 # FIRECRAWL_BASE_URL=https://api.firecrawl.dev
 # FIRECRAWL_TIMEOUT_MS=30000            # Per-request timeout (default: 30000 = 30s)

-# ── ChatGPT browser transport (Firefox-fingerprinted client) ──
-# Used by: open-sse/services/chatgptTlsClient.ts — native wreq-js request
-# timeout and the JS-side hard-deadline grace layered on top of it.
-# OMNIROUTE_CHATGPT_TLS_TIMEOUT_MS=60000
-# OMNIROUTE_CHATGPT_TLS_GRACE_MS=10000
-# Max wait for the FIRST streamed byte before switching from direct streaming
-# to a buffered response, in milliseconds. Default 30000 (30s). The request's
-# hard deadline continues to apply while the buffered body is read.
-# OMNIROUTE_CHATGPT_STREAM_FIRST_BYTE_TIMEOUT_MS=30000
-
-# ── Claude browser transport (Chromium-fingerprinted client) ──
-# Used by: open-sse/services/claudeTlsClient.ts — native wreq-js request
-# timeout and the JS-side hard-deadline grace layered on top of it.
+# ── Claude TLS sidecar (Chromium-fingerprinted client) ──
+# Used by: open-sse/services/claudeTlsClient.ts — wire-level timeout for
+# the bogdanfinn/tls-client koffi binding and the JS-side grace window
+# layered on top of it when the native library is wedged.
 # OMNIROUTE_CLAUDE_TLS_TIMEOUT_MS=60000
 # OMNIROUTE_CLAUDE_TLS_GRACE_MS=10000

-# ── Perplexity browser transport (Firefox-fingerprinted client) ──
-# Used by: open-sse/services/perplexityTlsClient.ts — native wreq-js request
-# timeout and the JS-side hard-deadline grace layered on top of it.
+# ── Perplexity TLS sidecar (Firefox-fingerprinted client) ──
+# Used by: open-sse/services/perplexityTlsClient.ts — wire-level timeout for
+# the bogdanfinn/tls-client koffi binding and the JS-side grace window
+# layered on top of it when the native library is wedged.
 # OMNIROUTE_PPLX_TLS_TIMEOUT_MS=30000
 # OMNIROUTE_PPLX_TLS_GRACE_MS=10000

@@ -1500,16 +1492,18 @@ CURSOR_USER_AGENT="Cursor/3.4"
 # meta-commentary. Set to 1/true/yes/on to restore the old behavior.
 # OMNIROUTE_PPLX_SEARCH_HINT=0

-# ── Grok web browser transport (Chrome-fingerprinted client) ──
-# Used by: open-sse/services/grokTlsClient.ts — native wreq-js request timeout
-# and the JS-side hard-deadline grace layered on top of it.
+# ── Grok web TLS sidecar (Chrome-fingerprinted client) ──
+# Used by: open-sse/services/grokTlsClient.ts — wire-level timeout for the
+# bogdanfinn/tls-client koffi binding and the JS-side grace window layered on
+# top of it when the native library is wedged.
 # OMNIROUTE_GROK_TLS_TIMEOUT_MS=60000
 # OMNIROUTE_GROK_TLS_GRACE_MS=10000

-# ── Notion web browser transport (Chrome-fingerprinted client) ──
-# Used by: open-sse/services/notionTlsClient.ts — native wreq-js request timeout
-# and the JS-side hard-deadline grace layered on top of it. The notion-web
-# executor raises the native timeout per-request to 180000 for long generations.
+# ── Notion web TLS sidecar (Chrome-fingerprinted client) ──
+# Used by: open-sse/services/notionTlsClient.ts — wire-level timeout for the
+# bogdanfinn/tls-client koffi binding and the JS-side grace window layered on
+# top of it when the native library is wedged. The notion-web executor raises
+# the wire timeout per-request to 180000 for long generations.
 # OMNIROUTE_NOTION_TLS_TIMEOUT_MS=30000
 # OMNIROUTE_NOTION_TLS_GRACE_MS=10000

--- a/.github/dependabot.yml
+++ b/.github/dependabot.yml
@@ -58,6 +58,22 @@ updates:
      # on the VPS — so keep auto-bumps frozen (no update-types = ignore every version).
      # Migrate it intentionally, not via dependabot (#4050).
      - dependency-name: "@huggingface/transformers"
+      # onnxruntime-node is the OTHER HALF of the @huggingface/transformers pair frozen
+      # above: the hoisted copy must equal the exact version transformers pins, or npm
+      # nests a second ABI-incompatible native copy (contract test
+      # tests/unit/onnxruntime-single-copy.test.ts, pair established in #9962). A solo
+      # bump can never be correct — it only ever moves together with transformers, in
+      # the same deliberate migration PR. Freezing it keeps the production group PRs
+      # (e.g. #12219) from being born red on the pair contract.
+      - dependency-name: "onnxruntime-node"
+      # eslint-plugin-react-hooks is pinned to 7.0.1 by a contract test
+      # (tests/unit/eslint-react-hooks-version-pinned.test.ts) until the 7.1.1 rule set
+      # is adopted deliberately — that adoption needs a full cold lint run and its own
+      # PR (the #12146 react-hooks migration finished on 2026-09-01, so the path is
+      # open; the bump still must not ride a dependabot group, where it reds the
+      # development group PRs, e.g. #12220). Remove this ignore in the adoption PR
+      # together with the pin test.
+      - dependency-name: "eslint-plugin-react-hooks"

  - package-ecosystem: "github-actions"
    directory: "/"
--- a/.github/workflows/docker-publish.yml
+++ b/.github/workflows/docker-publish.yml
@@ -535,7 +535,7 @@ jobs:
      - name: Upload Trivy SARIF to Security tab
        if: needs.prepare.outputs.version != 'main'
        continue-on-error: true
-        uses: github/codeql-action/upload-sarif@v4.37.8
+        uses: github/codeql-action/upload-sarif@v4.37.9
        with:
          sarif_file: trivy-results.sarif
          category: trivy-image
--- a/.github/workflows/electron-release.yml
+++ b/.github/workflows/electron-release.yml
@@ -237,7 +237,7 @@ jobs:
        #          optionals (@img/sharp-*, @img/sharp-libvips-*, @ngrok/ngrok-*,
        #          fsevents) carry linux forks. Replace them with the forks this
        #          leg's own `npm ci` resolved, then assert every bundled native
-        #          (better-sqlite3 prebuilds, wreq-js, onnxruntime)
+        #          (koffi triplets, better-sqlite3 prebuilds, wreq-js, onnxruntime)
        #          can service this leg's platform/arch before packaging starts.
        run: |
          node scripts/build/standaloneBundle.mjs restore --archive web-bundle.tar.gz
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -83,7 +83,7 @@ Client → /v1/chat/completions (Next.js route)

 API routes follow a consistent pattern: `Route → CORS preflight → Zod body validation → Optional auth (extractApiKey/isValidApiKey) → API key policy enforcement → Handler delegation (open-sse)`. No global Next.js middleware — interception is route-specific.

-**Combo routing** (`open-sse/services/combo.ts`): 19 public strategies (priority, weighted, fill-first, round-robin, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline). Each target calls `handleSingleModel()` which wraps `handleChatCore()` with per-target error handling and circuit breaker checks. The `fusion` strategy is the exception: it fans out to a panel of models in parallel, then a judge model synthesizes one final answer (`open-sse/services/fusion.ts`). See `docs/routing/AUTO-COMBO.md` for the 15-factor Auto-Combo scoring + the full strategy table and `docs/architecture/RESILIENCE_GUIDE.md` for the 3 resilience layers.
+**Combo routing** (`open-sse/services/combo.ts`): 19 public strategies (priority, weighted, fill-first, round-robin, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline). Each target calls `handleSingleModel()` which wraps `handleChatCore()` with per-target error handling and circuit breaker checks. The `fusion` strategy is the exception: it fans out to a panel of models in parallel, then a judge model synthesizes one final answer (`open-sse/services/fusion.ts`). See `docs/routing/AUTO-COMBO.md` for the 16-factor Auto-Combo scoring + the full strategy table and `docs/architecture/RESILIENCE_GUIDE.md` for the 3 resilience layers.

 ---

@@ -422,7 +422,7 @@ For any non-trivial change, read the matching deep-dive first:
 | Repo navigation                               | `docs/architecture/REPOSITORY_MAP.md`                   |
 | Architecture                                  | `docs/architecture/ARCHITECTURE.md`                     |
 | Engineering reference                         | `docs/architecture/CODEBASE_DOCUMENTATION.md`           |
-| Auto-Combo (15-factor scoring, 19 strategies) | `docs/routing/AUTO-COMBO.md`                            |
+| Auto-Combo (16-factor scoring, 19 strategies) | `docs/routing/AUTO-COMBO.md`                            |
 | Resilience (3 mechanisms)                     | `docs/architecture/RESILIENCE_GUIDE.md`                 |
 | Reasoning replay                              | `docs/routing/REASONING_REPLAY.md`                      |
 | Skills framework                              | `docs/frameworks/SKILLS.md`                             |
--- a/16
+++ b/16
@@ -103,11 +103,25 @@ RUN test -f package-lock.json \
 # node-gyp comes from npm's own bundled copy (deterministic, already in the image)
 # instead of `npx --yes`, which would install an arbitrary registry version
 # on-demand and run its lifecycle scripts (Sonar docker:S6505).
+#
+# tls-client-node (claude-web/grok-web/lmarena/perplexity-web TLS
+# impersonation) hits the same --ignore-scripts wall: its own postinstall.js
+# fetches a platform .so/.dylib/.dll from the bogdanfinn/tls-client GitHub
+# Releases API and is never invoked when npm ci skips lifecycle scripts. Unlike
+# better-sqlite3 above, that script never throws on failure — it only
+# `console.warn`s and exits 0 — so a rate-limited or offline build would
+# otherwise succeed silently with an empty bin/ and only fail at first request
+# in production (TlsClientUnavailableError, #7802). Run it explicitly here so
+# a broken/rate-limited fetch fails the BUILD loudly instead of shipping a
+# broken image.
 RUN --mount=type=cache,id=s/92ca8a61-c1ba-421f-a389-d48ac7258c2d-npm-cache,target=/root/.npm \
  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()"
+  && node -e "require('better-sqlite3')(':memory:').close()" \
+  && node node_modules/tls-client-node/scripts/postinstall.js \
+  && (test -n "$(find node_modules/tls-client-node/bin -mindepth 1 -print -quit 2>/dev/null)" \
+      || (echo "tls-client-node native binary missing after postinstall — GitHub API fetch likely rate-limited or failed (#7802)" >&2 && exit 1))

 # Build with Turbopack (stable in Next 16, the repo default). The v3.8.27-era
 # TurbopackInternalError panic ("entered unreachable code: there must be a path to a
--- a/Dockerfile.bun
+++ b/Dockerfile.bun
@@ -31,9 +31,9 @@ COPY scripts/dev/sync-env.mjs ./scripts/dev/sync-env.mjs
 # Fast Bun native package install
 RUN bun install --include=optional --quiet

-# Compile native better-sqlite3 Node-API addon under Bun
-RUN if [ -d "node_modules/better-sqlite3" ]; then \
-      (cd node_modules/better-sqlite3 && bunx node-gyp rebuild); \
+# Fetch tls-client-node native binary if script exists
+RUN if [ -f "node_modules/tls-client-node/scripts/postinstall.js" ] && [ ! -d "node_modules/tls-client-node/bin" ]; then \
+      bun node_modules/tls-client-node/scripts/postinstall.js || true; \
    fi

 # Smoke check native database driver used by Bun (bun:sqlite)
--- a/README.md
+++ b/README.md
@@ -49,11 +49,11 @@

 [![Discord](https://img.shields.io/badge/Discord-5865F2?style=for-the-badge&logo=discord&logoColor=white)](https://discord.gg/U47eFqAXCn)
 [![Telegram](https://img.shields.io/badge/Telegram-26A5E4?style=for-the-badge&logo=telegram&logoColor=white)](https://t.me/omnirouteOficial)
-[![WhatsApp Global](https://img.shields.io/badge/WhatsApp_Global-25D366?style=for-the-badge&logo=whatsapp&logoColor=white)](https://chat.whatsapp.com/JI7cDQ1GyaiDHhVBpLxf8b?mode=gi_t)
+[![WhatsApp Global](https://img.shields.io/badge/WhatsApp_Global-25D366?style=for-the-badge&logo=whatsapp&logoColor=white)](https://chat.whatsapp.com/FvuCbrpZmQ6I85n2vW5QIC?s=cl&p=a&mlu=4)
 [![WhatsApp Brasil](https://img.shields.io/badge/WhatsApp_Brasil-25D366?style=for-the-badge&logo=whatsapp&logoColor=white)](https://chat.whatsapp.com/KWgatljAjmbELQory59Oti?s=cl&p=a&mlu=4)
 [![Website](https://img.shields.io/badge/Website-omniroute.online-blue?logo=google-chrome&logoColor=white)](https://omniroute.online)

-**Questions, provider tips, roadmap & support → [Discord](https://discord.gg/U47eFqAXCn) · [Telegram](https://t.me/omnirouteOficial) · WhatsApp [🌍 Global](https://chat.whatsapp.com/JI7cDQ1GyaiDHhVBpLxf8b?mode=gi_t) / [🇧🇷 Brasil](https://chat.whatsapp.com/LTSpdFhXTxjH4R6CCNiKWz)**
+**Questions, provider tips, roadmap & support → [Discord](https://discord.gg/U47eFqAXCn) · [Telegram](https://t.me/omnirouteOficial) · WhatsApp [🌍 Global](https://chat.whatsapp.com/FvuCbrpZmQ6I85n2vW5QIC?s=cl&p=a&mlu=4) / [🇧🇷 Brasil](https://chat.whatsapp.com/KWgatljAjmbELQory59Oti?s=cl&p=a&mlu=4) / [Portal](https://portal.sthub.com.br/communities/groups/st-hub/channels/Omniroute-World-8kRjmK)**

 <br/>

@@ -431,7 +431,7 @@ All **19** strategies — mix & match per combo step:
  <tr>
    <td align="center">17</td>
    <td nowrap><code>auto</code></td>
-    <td>15-factor live scoring across every connection 🤖</td>
+    <td>16-factor live scoring across every connection 🤖</td>
  </tr>
  <tr>
    <td align="center">18</td>
@@ -445,7 +445,7 @@ All **19** strategies — mix & match per combo step:
  </tr>
 </table>

-<sub>The Auto-Combo engine scores every candidate on **15 factors** (health, quota, cost, latency, task fit, quality, session availability…) — see [`docs/routing/AUTO-COMBO.md`](docs/routing/AUTO-COMBO.md).</sub>
+<sub>The Auto-Combo engine scores every candidate on **16 factors** (health, quota, cost, latency, task fit, quality, session availability…) — see [`docs/routing/AUTO-COMBO.md`](docs/routing/AUTO-COMBO.md).</sub>

 ##

@@ -1183,10 +1183,10 @@ Métricas canônicas em 2026-08-24: **1.029 vídeos únicos** · **11.132.922 vi
 | 🐙 **GitHub** — follow for releases & tips | [@diegosouzapw](https://github.com/diegosouzapw)                                                                          |
 | 💬 **Discord**                             | [discord.gg/U47eFqAXCn](https://discord.gg/U47eFqAXCn)                                                                    |
 | ✈️ **Telegram**                            | [t.me/omnirouteOficial](https://t.me/omnirouteOficial)                                                                    |
-| 🟢 **WhatsApp — 🌍 Global**                | [join the group](https://chat.whatsapp.com/FvuCbrpZmQ6I85n2vW5QIC?s=cl&p=a&mlu=4)                                        |
+| 🟢 **WhatsApp — 🌍 Global**                | [join the group](https://chat.whatsapp.com/FvuCbrpZmQ6I85n2vW5QIC?s=cl&p=a&mlu=4)                                         |
 | 🟢 **WhatsApp — 🇧🇷 Brasil**                | [entrar no grupo](https://chat.whatsapp.com/KWgatljAjmbELQory59Oti?s=cl&p=a&mlu=4)                                        |
 | 🌍 **Website**                             | [omniroute.online](https://omniroute.online)                                                                              |
-| 🌍 **🌍StHub OmniRoute Community (free)** | [portal sthub](https://portal.sthub.com.br/communities/groups/st-hub/channels/Omniroute-World-8kRjmK)                     |
+| 🌍 **🌍StHub OmniRoute Community (free)**  | [portal sthub](https://portal.sthub.com.br/communities/groups/st-hub/channels/Omniroute-World-8kRjmK)                     |
 | 📦 **Source code**                         | [github.com/diegosouzapw/OmniRoute](https://github.com/diegosouzapw/OmniRoute)                                            |
 | 🐛 **Report a bug**                        | [open an issue](https://github.com/diegosouzapw/OmniRoute/issues) — attach `npm run system-info` output                   |
 | 🤝 **Contribute**                          | [CONTRIBUTING.md](CONTRIBUTING.md) · [Branching & Release Model](docs/ops/BRANCHING_MODEL.md) · pick a `good first issue` |
@@ -1269,7 +1269,7 @@ Métricas canônicas em 2026-08-24: **1.029 vídeos únicos** · **11.132.922 vi
  <tr><td nowrap><b><a href="docs/compression/COMPRESSION_RULES_FORMAT.md">Compression Rules Format</a></b></td><td>JSON rule-pack schemas for Caveman and RTK filters</td></tr>
  <tr><td nowrap><b><a href="docs/compression/COMPRESSION_LANGUAGE_PACKS.md">Compression Language Packs</a></b></td><td>Language detection and Caveman rule-pack authoring</td></tr>
  <tr><td nowrap><b><a href="docs/architecture/RESILIENCE_GUIDE.md">Resilience Guide</a></b></td><td>Circuit breakers, cooldowns, queue, anti-thundering herd, TLS spoofing</td></tr>
-  <tr><td nowrap><b><a href="docs/routing/AUTO-COMBO.md">Auto-Combo Engine</a></b></td><td>15-factor scoring, mode packs, self-healing</td></tr>
+  <tr><td nowrap><b><a href="docs/routing/AUTO-COMBO.md">Auto-Combo Engine</a></b></td><td>16-factor scoring, mode packs, self-healing</td></tr>
  <tr><td nowrap><b><a href="docs/ops/PROXY_GUIDE.md">Proxy Guide</a></b></td><td>3-level proxy system, 1proxy marketplace, registry CRUD</td></tr>
  <tr><td nowrap><b><a href="docs/reference/FREE_TIERS.md">Free Tiers</a></b></td><td>Consolidated directory: 38 documented recurring pools / 446 cataloged free-tier entries</td></tr>
  <tr><td nowrap><b><a href="docs/guides/FEATURES.md">Features Gallery</a></b></td><td>Visual dashboard tour with screenshots</td></tr>
--- a/THIRD_PARTY_NOTICES.md
+++ b/THIRD_PARTY_NOTICES.md
@@ -25,31 +25,6 @@ NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FO
 DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT
 OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

-## wreq-js 3.0.0
-
-OmniRoute distributes `wreq-js` and its seven platform-specific native addons from
-[`wreq-js@3.0.0`](https://www.npmjs.com/package/wreq-js/v/3.0.0).
-
-MIT License
-
-Copyright (c) 2025 will-work-for-meal
-Copyright (c) 2025 Oleksandr Herasymov
-
-Permission is hereby granted, free of charge, to any person obtaining a copy of this software and
-associated documentation files (the "Software"), to deal in the Software without restriction,
-including without limitation the rights to use, copy, modify, merge, publish, distribute,
-sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT
-NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
-NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT
-OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
 ## blackwell-systems/gcf-typescript

 The generic-profile codec in
--- a/changelog.d/features/12218-rankings-reliability-sort.md
+++ b/changelog.d/features/12218-rankings-reliability-sort.md
@@ -0,0 +1 @@
+- **feat(rankings):** order Free Provider Rankings by what each provider actually served — `GET /api/free-provider-rankings?sortBy=reliability` and a "Most reliable first" toggle on the page. Providers with too few calls to state a success rate keep their score order below the measured ones; the default order is unchanged ([#12218](https://github.com/diegosouzapw/OmniRoute/pull/12218)).
--- a/changelog.d/features/12317-reliability-scoring-factor.md
+++ b/changelog.d/features/12317-reliability-scoring-factor.md
@@ -0,0 +1 @@
+- **feat(auto-combo):** Auto-Combo scoring can now weigh how often a provider/model has actually succeeded. The engine already carried that number on every candidate — 24 hours of usage history behind a ten-sample floor, real-time metrics otherwise — and the scoring function never read it, while the weight table described `stability` as if it did. `reliability` (`1 - failureRate`, with the same field precedence and the same rate-bounding the speed ranking already uses, so a corrupt reading means "nothing observed" rather than "fails every call") is now a declared factor shipping at weight `0`, so routing is unchanged until an operator gives it one, and the `stability` description now matches what that factor computes ([#12317](https://github.com/diegosouzapw/OmniRoute/pull/12317))
--- a/changelog.d/features/12319-strict-zero-cost-exclusion-reason.md
+++ b/changelog.d/features/12319-strict-zero-cost-exclusion-reason.md
@@ -0,0 +1 @@
+- **feat(routing):** With `freeAccessPolicy: "strict"`, the read-only candidate listing (`GET /v1/auto-combo/{channel}/candidates`) no longer hides the candidates the zero-cost guard excludes — the same read-only transparency the resilience filter already honours (#9133). Each candidate now carries `freeAccessExclusion` saying why it would be kept out, and it tells an exhausted allowance apart from a quota reading that never arrived or went stale, which used to look identical from the outside. Routing is unchanged: the listing reports, it never enforces. The separate `excludeTosAvoid` guard still drops its candidates without a reason; that gap is now documented rather than closed ([#12319](https://github.com/diegosouzapw/OmniRoute/pull/12319))
--- a/changelog.d/features/12320-radar-catalog-limits-training.md
+++ b/changelog.d/features/12320-radar-catalog-limits-training.md
@@ -0,0 +1 @@
+- **feat(radar):** The Radar catalog table now shows two facts it was already receiving from the feed and dropping on the floor: the per-model rate limits (requests and tokens, per minute and per day) in a new column, and a badge when a provider's terms state it may train on the prompts you send. A limit of zero renders as zero rather than "rate-only" — for a ceiling those are opposite facts — and a model with no training statement gets no badge, because an absent statement is not a guarantee ([#12320](https://github.com/diegosouzapw/OmniRoute/pull/12320))
--- a/changelog.d/fixes/12151-passthrough-usage-estimation.md
+++ b/changelog.d/fixes/12151-passthrough-usage-estimation.md
@@ -0,0 +1 @@
+- **fix(sse):** passthrough streams now estimate usage on finish when upstream closes without usage even with `stream_options.include_usage` — avoids `0 tokens / 0%` for providers that stay silent (and correctly handles trailing empty-choices usage) ([#12151](https://github.com/diegosouzapw/OmniRoute/pull/12151))
--- a/changelog.d/fixes/12177-combo-clear-agent-features.md
+++ b/changelog.d/fixes/12177-combo-clear-agent-features.md
@@ -0,0 +1,4 @@
+- **fix(combos):** clearing an agent feature in the combos editor now persists — unchecking
+  context cache protection, or emptying the system message or tool filter, sends an explicit
+  `null` instead of dropping the field from the `PUT` body, which the update merge read as
+  "leave unchanged" ([#12177](https://github.com/diegosouzapw/OmniRoute/pull/12177)) — thanks @foreveryh
--- a/changelog.d/fixes/12215-free-tier-overlay-freshness.md
+++ b/changelog.d/fixes/12215-free-tier-overlay-freshness.md
@@ -0,0 +1 @@
+- **fix(free-tier):** `/api/free-tier/summary` no longer computes its totals from a Radar feed built before the catalog the running release ships. When the cached feed is older — or carries no build date at all — the route answers from the shipped catalog, resolved through the operator's local model state so disabled and tombstoned models stay out of the numbers ([#12215](https://github.com/diegosouzapw/OmniRoute/pull/12215)).
--- a/changelog.d/fixes/12242-passthrough-402-per-model-lockout.md
+++ b/changelog.d/fixes/12242-passthrough-402-per-model-lockout.md
@@ -0,0 +1 @@
+- **fix(resilience):** a 402 on a single paid model of a passthrough/gateway provider (e.g. `kilo-gateway`, `ollama-cloud`) no longer terminalizes the whole connection with a never-auto-recovered `credits_exhausted` status — only the paid model is locked out, so free models on the same key keep serving. 402 variant of [#3027](https://github.com/diegosouzapw/OmniRoute/issues/3027). Single-credential providers are unaffected — a 402 there is still treated as the key being genuinely out of credit ([#5239](https://github.com/diegosouzapw/OmniRoute/issues/5239) / [#10616](https://github.com/diegosouzapw/OmniRoute/issues/10616)) ([#12242](https://github.com/diegosouzapw/OmniRoute/issues/12242)) — thanks @brick30llc-ctrl
--- a/changelog.d/fixes/12314-scoring-weights-coverage.md
+++ b/changelog.d/fixes/12314-scoring-weights-coverage.md
@@ -0,0 +1 @@
+- **fix(combo):** Expose the two Auto-Combo scoring factors nobody could set — the combo validation schema and the dashboard weight sliders both declared 13 of the scorer's 15 factors, so `connectionDensity` (spreads load across a provider's connections) and `quality` were dropped on save and offered nowhere. The sliders also shipped their own default table that differed from the engine's on every non-zero factor and summed to 1.05, so the percentages shown next to them added up to 105%. Both lists now match `DEFAULT_WEIGHTS`, and a test keeps them there. Note that a combo whose stored `weights` omitted the two keys was effectively running with them at zero and the other thirteen renormalized upward; it now runs with the engine's intended distribution, so its routing does shift ([#12314](https://github.com/diegosouzapw/OmniRoute/pull/12314))
--- a/changelog.d/fixes/12318-free-tiers-confidence-claim.md
+++ b/changelog.d/fixes/12318-free-tiers-confidence-claim.md
@@ -0,0 +1 @@
+- **fix(docs):** The free-tier reference no longer says its numbers come "confidence tagged per row" — no catalog entry carries a confidence tag and the API serves none, so every figure on that page is an estimate of the same, unstated quality. The page now states what an entry does vouch for: an independently documented hard stop (set by hand with the source in a comment, never defaulted to `true`) and a prompt-training disclosure, both with live counts the `check:docs-counts` gate keeps honest ([#12318](https://github.com/diegosouzapw/OmniRoute/pull/12318))
--- a/changelog.d/fixes/12321-usage-fetcher-registration.md
+++ b/changelog.d/fixes/12321-usage-fetcher-registration.md
@@ -0,0 +1 @@
+- **fix(usage):** `adobe-firefly` and `firefly` have had a working usage fetcher since Adobe Firefly landed, but neither was ever added to the registration list, so the provider-plugin manifest, `genericQuotaFetcher` and the free-access quota cache all reported them as having no usage support — while `USAGE_SUPPORTED_PROVIDERS` said the opposite. Both are now declared, which also means their credit balance is fetched like any other declared provider's: `registerGenericQuotaFetchers` now registers a generic quota fetcher for them, and `resolveFreeAccessState` no longer returns early. A test holds the registration list to the dispatcher's switch in both directions, which is what the module's own docstring already asked for in prose ([#12321](https://github.com/diegosouzapw/OmniRoute/pull/12321))
--- a/changelog.d/maintenance/11753-wreq-js-transport.md
+++ b/changelog.d/maintenance/11753-wreq-js-transport.md
@@ -1 +0,0 @@
- **chore(stealth):** replace the `tls-client-node` sidecar/temp-file transport used by the six web-cookie providers with the exactly pinned `wreq-js` 3.0.0 native transport, preserving streaming, proxy isolation, deadlines, EOF policies, binary responses, and cancellation while removing the obsolete downloader and native repair path ([#11753](https://github.com/diegosouzapw/OmniRoute/pull/11753)).
--- a/changelog.d/maintenance/12316-mode-packs-doc-gate.md
+++ b/changelog.d/maintenance/12316-mode-packs-doc-gate.md
@@ -0,0 +1 @@
+- **docs(auto-combo):** The mode pack table in `docs/routing/AUTO-COMBO.md` now lists all six shipped packs with every weight each one sets, replacing a four-pack table whose numbers had also drifted from the source. It states plainly that no pack sets `quality`, so selecting any pack silences the observed-quality signal. Six more documents that quote the scoring factor count joined the `check:docs-counts` gate, which caught five stale claims — including one naming nine factors that do not exist — and two stale mode pack counts. The dashboard routing panel, which offered four of the six packs and labelled the default strategy "6-Factor Scoring", is now covered by a test; the two packs it was missing are `reliability-first` and `chaos-mode`, the latter labelled as the fault-injection profile it is rather than as one more routing preference ([#12316](https://github.com/diegosouzapw/OmniRoute/pull/12316))
--- a/config/quality/.license-allowlist.json
+++ b/config/quality/.license-allowlist.json
@@ -74,6 +74,12 @@
      "justification": "CC-BY-4.0 applies to the caniuse browser-support data (a dataset, not code). The Creative Commons Attribution license requires attribution when distributing — OmniRoute does not distribute caniuse-lite data directly to end users; it is consumed by browserslist/PostCSS at build time to generate CSS compatibility info. This is a widely accepted pattern in the Node.js ecosystem (caniuse-lite is in millions of projects). Attribution is satisfied by keeping the package in node_modules with its original license file.",
      "risk": "low",
      "reviewAt": "v4.0.0"
+    },
+    "tls-client-node": {
+      "license": "Custom: LICENSE (Apache-2.0 + Commons Clause)",
+      "justification": "TODO: revisar — tls-client-node uses Apache-2.0 with a 'Commons Clause' addendum that restricts 'Selling' the software (i.e., offering it as a hosted/commercial service whose value derives substantially from tls-client-node). OmniRoute is an open-source proxy; however if deployed as a paid SaaS/hosting service, this restriction could apply. The package is used by grokTlsClient.ts for Grok TLS fingerprinting. RISK: medium — legal review recommended before commercial deployment. Alternatives: consider replacing with a native TLS fingerprinting approach or a truly permissive library.",
+      "risk": "medium",
+      "reviewAt": "v3.9.0"
    }
  }
 }
--- a/config/quality/dependency-allowlist.json
+++ b/config/quality/dependency-allowlist.json
@@ -139,6 +139,7 @@
    "tailwind-merge",
    "tailwindcss",
    "tiktoken",
+    "tls-client-node",
    "tsup",
    "tsx",
    "turndown",
--- a/config/release/wreq-js-native-manifest.json
+++ b/config/release/wreq-js-native-manifest.json
@@ -1,44 +0,0 @@
-{
-  "package": "wreq-js",
-  "version": "3.0.0",
-  "source": "https://registry.npmjs.org/wreq-js/-/wreq-js-3.0.0.tgz",
-  "npmIntegrity": "sha512-RZCoRSevVPpH4A4B4MxbFGo/pVPFveWd2gbe4ENKpPWlKXEYklZSDESOjBMmrIsmnkHh+nhM4PNJvG+NL7wBPA==",
-  "license": "MIT",
-  "nativeAddons": [
-    {
-      "path": "rust/wreq-js.darwin-arm64.node",
-      "size": 7690880,
-      "sha256": "c82eec39df691adb94f2cd09a8ff51335de8587cf132cd8b3ec797469a4b5002"
-    },
-    {
-      "path": "rust/wreq-js.darwin-x64.node",
-      "size": 8192028,
-      "sha256": "073b8a8a4c26aedbce7c14eef3e5567918e62e8dbf4d28296b23f9d2beec2981"
-    },
-    {
-      "path": "rust/wreq-js.linux-arm64-gnu.node",
-      "size": 8520824,
-      "sha256": "861d96a78caf7ce02c9ae8d37f1c59f5b0480e3142775c32917fcfe9b88524b0"
-    },
-    {
-      "path": "rust/wreq-js.linux-arm64-musl.node",
-      "size": 8735472,
-      "sha256": "2409a3578c8c440df419b4d5abe3ac149bec48881611a6dc1571b95e6246552d"
-    },
-    {
-      "path": "rust/wreq-js.linux-x64-gnu.node",
-      "size": 9048992,
-      "sha256": "55b40f4602c52111dfcdcc93db83f9d0de55d0ef7540348757709d58d05a9b64"
-    },
-    {
-      "path": "rust/wreq-js.linux-x64-musl.node",
-      "size": 8974880,
-      "sha256": "bd52d15b1bb4704b11561a8aa95648a6c91150082b5af0e39dd1608b7db2d317"
-    },
-    {
-      "path": "rust/wreq-js.win32-x64-msvc.node",
-      "size": 7967232,
-      "sha256": "7451a8701b82c946b03ba2be2f15257260a250b9e0ed9910611b22564fbec7a9"
-    }
-  ]
-}
--- a/docs/architecture/ARCHITECTURE.md
+++ b/docs/architecture/ARCHITECTURE.md
@@ -370,14 +370,18 @@ Key capabilities:
  **auto**, lkgp, context-optimized, context-relay, **fusion**, plus a fallback path) —
  auto is the headline addition in v3.8.0; `fusion` (panel fan-out + judge synthesis,
  `open-sse/services/fusion.ts`) is new in v3.8.36.
- **9-factor scoring**: cost, latency p95, success rate, quota headroom, lockout
-  proximity, breaker state, recent failures, model availability, and tag affinity.
+- **16-factor scoring**: quota, health, inverse cost, inverse latency, task fit and
+  ten more. The canonical table of factors and their default weights lives in
+  [`docs/routing/AUTO-COMBO.md`](../routing/AUTO-COMBO.md) — restating it here would
+  give it a second place to go stale.
 - **Virtual factory** materializes ephemeral combos when no matching named combo
  exists, sourcing candidates from healthy active provider connections.
 - **Auto prefixes**: `auto/coding`, `auto/cheap`, `auto/fast`, `auto/offline`,
  `auto/smart`, `auto/lkgp` — each backed by a tuned weight profile.
- **4 mode packs**: coding, fast, cheap, smart — shipped as preset weight
-  configurations callable from the dashboard.
+- **6 mode packs**: `ship-fast`, `cost-saver`, `quality-first`, `offline-friendly`,
+  `reliability-first` and `chaos-mode` — preset weight configurations callable from
+  the dashboard. (Not to be confused with the `auto/*` prefixes above, which are
+  request-time variants.)

 For full algorithmic detail (factor formulas, weight tuning), see
 [`docs/routing/AUTO-COMBO.md`](../routing/AUTO-COMBO.md).
--- a/docs/architecture/REPOSITORY_MAP.md
+++ b/docs/architecture/REPOSITORY_MAP.md
@@ -402,24 +402,24 @@ open-sse/

 ### Subsystem deep-dives

-| Doc                        | Purpose                                                             |
-| -------------------------- | ------------------------------------------------------------------- |
-| `MCP-SERVER.md`            | MCP server: 110 tools, 3 transports, 33 scopes, REST endpoints      |
-| `A2A-SERVER.md`            | A2A v0.3: JSON-RPC, 6 skills, REST helpers, agent card              |
-| `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                         |
-| `COMPLIANCE.md`            | Audit log, retention, noLog opt-out                                 |
-| `WEBHOOKS.md`              | HMAC-signed webhook delivery                                        |
-| `REASONING_REPLAY.md`      | Hybrid memory/SQLite cache for `reasoning_content`                  |
-| `AUTHZ_GUIDE.md`           | Authorization pipeline (`classify` → `policies` → `enforce`)        |
-| `RESILIENCE_GUIDE.md`      | Circuit breaker + cooldown + model lockout                          |
-| `STEALTH_GUIDE.md`         | TLS fingerprinting (JA3/JA4), Claude Code CCH, MITM cert            |
-| `AUTO-COMBO.md`            | Auto Combo engine (9-factor scoring, 4 mode packs, virtual factory) |
+| Doc                        | Purpose                                                              |
+| -------------------------- | -------------------------------------------------------------------- |
+| `MCP-SERVER.md`            | MCP server: 110 tools, 3 transports, 33 scopes, REST endpoints       |
+| `A2A-SERVER.md`            | A2A v0.3: JSON-RPC, 6 skills, REST helpers, agent card               |
+| `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                          |
+| `COMPLIANCE.md`            | Audit log, retention, noLog opt-out                                  |
+| `WEBHOOKS.md`              | HMAC-signed webhook delivery                                         |
+| `REASONING_REPLAY.md`      | Hybrid memory/SQLite cache for `reasoning_content`                   |
+| `AUTHZ_GUIDE.md`           | Authorization pipeline (`classify` → `policies` → `enforce`)         |
+| `RESILIENCE_GUIDE.md`      | Circuit breaker + cooldown + model lockout                           |
+| `STEALTH_GUIDE.md`         | TLS fingerprinting (JA3/JA4), Claude Code CCH, MITM cert             |
+| `AUTO-COMBO.md`            | Auto Combo engine (16-factor scoring, 6 mode packs, virtual factory) |

 ### Compression

--- a/docs/architecture/RESILIENCE_GUIDE.md
+++ b/docs/architecture/RESILIENCE_GUIDE.md
@@ -652,4 +652,4 @@ default `test:integration`, chaos and heap self-skip (without `RUN_CHAOS_INT`/`-

 - [Architecture Guide](./ARCHITECTURE.md) — System architecture and internals
 - [User Guide](../guides/USER_GUIDE.md) — Providers, combos, CLI integration
- [Auto-Combo Engine](../routing/AUTO-COMBO.md) — 13-factor scoring, mode packs
+- [Auto-Combo Engine](../routing/AUTO-COMBO.md) — 16-factor scoring, mode packs
--- a/docs/diagrams/auto-combo-scoring.mmd
+++ b/docs/diagrams/auto-combo-scoring.mmd
@@ -1,13 +1,13 @@
-%% Auto-Combo 15-factor scoring
+%% Auto-Combo 16-factor scoring
 %% Reflects: open-sse/services/autoCombo/scoring.ts (DEFAULT_WEIGHTS, sum = 1.0)
 %% v3.8.50
-%% svg-title: OmniRoute Auto-Combo 15-factor scoring
+%% svg-title: OmniRoute Auto-Combo 16-factor scoring
 %% svg-description: Flow from an incoming request through eligible candidates, the 15 weighted scoring factors, descending score sort, top-N selection, and sequential dispatch.
 flowchart TB
  Request["Incoming request"] --> Candidates["Eligible candidates<br/>(provider × model × account)"]
  Candidates --> Score["Compute composite score<br/>per candidate"]

-  subgraph Factors["15-factor scoring weights (sum = 1.0)"]
+  subgraph Factors["16-factor scoring weights (sum = 1.0)"]
    f1["quota (0.1429)"]
    f2["health (0.1605)"]
    f3["costInv (0.1429)"]
--- a/docs/diagrams/strategies-grid.svg
+++ b/docs/diagrams/strategies-grid.svg
@@ -95,7 +95,7 @@
 <rect width="173" height="158" rx="10" fill="#161b22" stroke="#ffffff" stroke-opacity="0.07" stroke-width="1"/>
 <text x="12" y="21" font-family="Consolas, 'Courier New', monospace" font-size="11" fill="#a78bfa">auto</text>
 <circle cx="20" cy="79" r="4" fill="none" stroke="#c9d1d9" stroke-width="1.6"/><circle cx="20" cy="79" r="1.6" fill="#c9d1d9"/><path d="M 26,79 C 62,79 84,67.5 112,67.5" fill="none" stroke="#8b5cf6" stroke-opacity="0.55" stroke-width="1.6"/><rect x="116" y="38.0" width="26" height="11" rx="2.5" fill="#1c2330" stroke="#ffffff" stroke-opacity="0.10" stroke-width="1"/><text x="147" y="46.5" font-family="Consolas, 'Courier New', monospace" font-size="8.5" fill="#71717a">72</text><rect x="116" y="62.0" width="26" height="11" rx="2.5" fill="#1c2330" stroke="#7ee787" stroke-opacity="0.8" stroke-width="1"/><text x="147" y="70.5" font-family="Consolas, 'Courier New', monospace" font-size="8.5" fill="#71717a">91</text><rect x="116" y="86.0" width="26" height="11" rx="2.5" fill="#1c2330" stroke="#ffffff" stroke-opacity="0.10" stroke-width="1"/><text x="147" y="94.5" font-family="Consolas, 'Courier New', monospace" font-size="8.5" fill="#71717a">64</text><rect x="116" y="110.0" width="26" height="11" rx="2.5" fill="#1c2330" stroke="#ffffff" stroke-opacity="0.10" stroke-width="1"/><text x="147" y="118.5" font-family="Consolas, 'Courier New', monospace" font-size="8.5" fill="#71717a">55</text><circle r="2.8" fill="#a78bfa" opacity="0"><animateMotion path="M 26,79 C 62,79 84,67.5 110,67.5" begin="3.3s" dur="3.6s" repeatCount="indefinite"/><animate attributeName="opacity" values="0;1;1;0;0" keyTimes="0;0.02;0.3;0.33999999999999997;1" begin="3.3s" dur="3.6s" repeatCount="indefinite"/></circle>
-<text x="12" y="148" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="9.5" fill="#71717a">live 15-factor scoring</text>
+<text x="12" y="148" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="9.5" fill="#71717a">live 16-factor scoring</text>
 </g><g transform="translate(796,456)">
 <rect width="173" height="158" rx="10" fill="#161b22" stroke="#ffffff" stroke-opacity="0.07" stroke-width="1"/>
 <text x="12" y="21" font-family="Consolas, 'Courier New', monospace" font-size="11" fill="#a78bfa">fusion</text>
--- a/docs/frameworks/OPEN_SSE_ARCHITECTURE.md
+++ b/docs/frameworks/OPEN_SSE_ARCHITECTURE.md
@@ -105,7 +105,7 @@ Resolves the request to a concrete `(provider, model, account, credentials)` tup

 For `auto/*` models, this stage also:

- Runs the **15-factor scoring** algorithm (`services/autoCombo/`)
+- Runs the **16-factor scoring** algorithm (`services/autoCombo/`)
 - Selects a `provider+model` pair based on health, cost, latency, etc.

 ### Stage 2: Translate (translator/)
@@ -245,7 +245,7 @@ Supports **19 routing strategies** (see `src/shared/constants/routingStrategies.
 | `reset-window`      | Reset window-based routing                                                |
 | `headroom`          | Most remaining quota headroom first                                       |
 | `strict-random`     | Truly uniform (no quality weighting)                                      |
-| `auto`              | Use 15-factor scoring (`autoCombo/`)                                      |
+| `auto`              | Use 16-factor scoring (`autoCombo/`)                                      |
 | `lkgp`              | Last known good provider first                                            |
 | `context-optimized` | Best for long-context requests                                            |
 | `fusion`            | Fan out to a panel in parallel, then synthesize via a judge (`fusion.ts`) |
@@ -280,7 +280,7 @@ Services are **focused, single-purpose modules** that handlers compose. The big
 ### Routing & Combo

 - `combo.ts` — entry point for combo-routed requests
- `services/autoCombo/` — 15-factor scoring, 8 auto routing strategies
+- `services/autoCombo/` — 16-factor scoring, 8 auto routing strategies
 - `wildcardRouter.ts` — matches wildcard routes (`gpt-*`)
 - `modelFamilyFallback.ts` — T5 intra-family fallback

@@ -570,7 +570,7 @@ The routing engine has strict performance budgets:
 - [ARCHITECTURE.md](../architecture/ARCHITECTURE.md) — high-level architecture
 - [CODEBASE_DOCUMENTATION.md](../architecture/CODEBASE_DOCUMENTATION.md) — engineering reference
 - [REPOSITORY_MAP.md](../architecture/REPOSITORY_MAP.md) — directory-by-directory
- [AUTO-COMBO.md](../routing/AUTO-COMBO.md) — 15-factor scoring
+- [AUTO-COMBO.md](../routing/AUTO-COMBO.md) — 16-factor scoring
 - [MCP-SERVER.md](./MCP-SERVER.md) — MCP server
 - [A2A-SERVER.md](./A2A-SERVER.md) — A2A server
 - Source: `open-sse/` (400+ files, ~143K LOC)
--- a/docs/getting-started/AUTO-COMBO-GUIDE.md
+++ b/docs/getting-started/AUTO-COMBO-GUIDE.md
@@ -46,14 +46,14 @@ model: "auto/cheap"     # Cheapest option

 ## Which "auto" Should I Use?

-| If you want... | Use this | Best for | How it works |
-|----------------|----------|----------|--------------|
-| **Best overall** | `auto` | General questions, chat | Balances speed, cost, and quality |
-| **Best code** | `auto/coding` | Writing code, debugging | Picks models good at coding tasks |
-| **Fastest response** | `auto/fast` | Quick answers, low latency | Prioritizes speed over everything |
-| **Cheapest option** | `auto/cheap` | Saving money | Picks the cheapest provider |
-| **Smartest model** | `auto/smart` | Complex tasks | Quality-first + explores new models |
-| **Most available** | `auto/offline` | When providers are busy | Picks providers with most capacity |
+| If you want...       | Use this       | Best for                   | How it works                        |
+| -------------------- | -------------- | -------------------------- | ----------------------------------- |
+| **Best overall**     | `auto`         | General questions, chat    | Balances speed, cost, and quality   |
+| **Best code**        | `auto/coding`  | Writing code, debugging    | Picks models good at coding tasks   |
+| **Fastest response** | `auto/fast`    | Quick answers, low latency | Prioritizes speed over everything   |
+| **Cheapest option**  | `auto/cheap`   | Saving money               | Picks the cheapest provider         |
+| **Smartest model**   | `auto/smart`   | Complex tasks              | Quality-first + explores new models |
+| **Most available**   | `auto/offline` | When providers are busy    | Picks providers with most capacity  |

 ### Examples

@@ -81,7 +81,7 @@ curl http://localhost:20128/v1/chat/completions \
 When you send a request with `model: "auto"`, OmniRoute:

 1. **Looks at all your connected providers** — Every provider you've added (OpenAI, Anthropic, Google, etc.)
-2. **Scores each one** on 5 factors:
+2. **Scores each one**, weighing among other things:
   - Is it working? (health)
   - Does it have capacity? (quota)
   - How much does it cost? (price)
@@ -94,29 +94,29 @@ When you send a request with `model: "auto"`, OmniRoute:

 Each provider gets a score from 0 to 1. The higher the score, the better the fit.

-| Factor | Weight | What it means |
-|--------|--------|---------------|
-| Health | 20% | Is the provider working? (circuit breaker state) |
-| Quota | 15% | Does it have capacity remaining? |
-| Cost | 15% | How expensive is it? (cheaper = higher score) |
-| Speed | 12% | How fast is it? (lower latency = higher score) |
-| Task Fit | 8% | Is it good at this type of task? |
-| Stability | 5% | Is it consistent? (low error rate) |
-| Tier | 5% | Account tier (Ultra > Pro > Free) |
-| Other | 20% | Context affinity, connection density, etc. |
+| Factor    | Weight | What it means                                    |
+| --------- | ------ | ------------------------------------------------ |
+| Health    | 20%    | Is the provider working? (circuit breaker state) |
+| Quota     | 15%    | Does it have capacity remaining?                 |
+| Cost      | 15%    | How expensive is it? (cheaper = higher score)    |
+| Speed     | 12%    | How fast is it? (lower latency = higher score)   |
+| Task Fit  | 8%     | Is it good at this type of task?                 |
+| Stability | 5%     | Is it consistent? (low error rate)               |
+| Tier      | 5%     | Account tier (Ultra > Pro > Free)                |
+| Other     | 20%    | Context affinity, connection density, etc.       |

 ### How Variants Change the Scoring

 Each variant uses different weights:

-| Variant | Prioritizes | Key Weights |
-|---------|-------------|-------------|
-| `auto` | Balanced | health=20%, quota=15%, cost=15% |
-| `auto/coding` | Quality | taskFit=37%, stability=15% |
-| `auto/fast` | Speed | latency=32%, health=28% |
-| `auto/cheap` | Cost | cost=37% |
-| `auto/smart` | Quality + Explore | taskFit=37%, exploration=10% |
-| `auto/offline` | Capacity | quota=37%, health=28% |
+| Variant        | Prioritizes       | Key Weights                     |
+| -------------- | ----------------- | ------------------------------- |
+| `auto`         | Balanced          | health=20%, quota=15%, cost=15% |
+| `auto/coding`  | Quality           | taskFit=37%, stability=15%      |
+| `auto/fast`    | Speed             | latency=32%, health=28%         |
+| `auto/cheap`   | Cost              | cost=37%                        |
+| `auto/smart`   | Quality + Explore | taskFit=37%, exploration=10%    |
+| `auto/offline` | Capacity          | quota=37%, health=28%           |

 ---

@@ -125,15 +125,19 @@ Each variant uses different weights:
 OmniRoute has **three layers of protection**:

 ### 1. Auto-Fallback
+
 If the best provider fails, OmniRoute automatically tries the next one. You don't need to do anything.

 ### 2. Self-Healing
+
 If a provider keeps failing:
+
 - **Score < 0.2** → Excluded for 5 minutes
 - **Circuit breaker open** → Auto-excluded
 - **More than 50% providers down** → Incident mode (no exploration)

 ### 3. Emergency Fallback
+
 If all providers fail, OmniRoute routes to stable free providers (like Kiro or Qoder) as a last resort.

 ---
@@ -209,7 +213,8 @@ Round-robin cycles through providers in order. Auto-combo **scores each provider
 ## Learn More

 For developers and contributors, see the [Auto-Combo Technical Reference](../routing/AUTO-COMBO.md) for:
- Full 13-factor scoring algorithm
+
+- Full 16-factor scoring algorithm
 - Mode pack weight tables
 - Implementation file paths
 - API endpoints
--- a/docs/guides/FEATURES.md
+++ b/docs/guides/FEATURES.md
@@ -18,7 +18,7 @@ Visual guide to every section of the OmniRoute dashboard.

 The v3.7.x → v3.8.0 cycle added zero-config auto routing, new providers, OAuth flows, deeper resilience, and a much richer CLI experience. Headline features below — full details further in the document and in linked specs.

- 🤖 **Auto Combo / Zero-config auto-routing** — use prefixes `auto/coding`, `auto/fast`, `auto/cheap`, `auto/offline`, `auto/smart`, `auto/lkgp`, `auto/chaos`. Backed by a 15-factor scoring engine and 6 curated **mode packs** (ship-fast, cost-saver, quality-first, offline-friendly)
+- 🤖 **Auto Combo / Zero-config auto-routing** — use prefixes `auto/coding`, `auto/fast`, `auto/cheap`, `auto/offline`, `auto/smart`, `auto/lkgp`, `auto/chaos`. Backed by a 16-factor scoring engine and 6 curated **mode packs** (ship-fast, cost-saver, quality-first, offline-friendly, reliability-first, chaos-mode)
 - 🆕 **Command Code provider** (#2199) — first-class registration with model catalog and quota tracking
 - 🆕 **Z.AI provider** — new free-tier provider with quota labels
 - 🎬 **KIE media expansion** — extended catalog including video generation models
--- a/docs/guides/FREE_PROVIDER_RANKINGS.md
+++ b/docs/guides/FREE_PROVIDER_RANKINGS.md
@@ -238,7 +238,7 @@ provider that ships at least one strong free model.
   `user_override → arena_elo → models_dev_tier → static table`). So after you connect
   the top free providers, routing with `model: "auto"` (e.g. `auto/coding`) will
   automatically prefer the higher-quality free models per request. See
-   [Auto-Combo](../routing/AUTO-COMBO.md) for the full 15-factor scoring.
+   [Auto-Combo](../routing/AUTO-COMBO.md) for the full 16-factor scoring.

 ---

@@ -248,7 +248,7 @@ provider that ships at least one strong free model.
  providers, no credit card required.
 - [Free Tiers directory](../reference/FREE_TIERS.md) — full catalog of free providers
  and their limits.
- [Auto-Combo](../routing/AUTO-COMBO.md) — the 15-factor routing engine that consumes the
+- [Auto-Combo](../routing/AUTO-COMBO.md) — the 16-factor routing engine that consumes the
  same Arena ELO task-fitness data.
 - [Environment variables](../reference/ENVIRONMENT.md) — `ARENA_ELO_SYNC_ENABLED` /
  `ARENA_ELO_SYNC_INTERVAL` reference.
--- a/docs/guides/TIERS.md
+++ b/docs/guides/TIERS.md
@@ -69,7 +69,7 @@ rate limits — circuit breaker recovers them on backoff.
 Dashboard → **Tiers** → assign your providers. Defaults (from `tierDefaults.json`) are
 sensible; edit when you have specific subscriptions to prioritize or providers to exclude.

-Auto-Combo's 15-factor scoring also considers tier. See
+Auto-Combo's 16-factor scoring also considers tier. See
 [`docs/routing/AUTO-COMBO.md`](../routing/AUTO-COMBO.md).

 ## Telemetry
--- a/docs/guides/TROUBLESHOOTING.md
+++ b/docs/guides/TROUBLESHOOTING.md
@@ -75,7 +75,7 @@ When you run `npm install -g omniroute`, you may see a wall of warnings like `np
 The warnings come from stale peer-dependency ranges in third-party packages OmniRoute doesn't control:

 1. **`marked-terminal` wants `marked >=1 <16`, found `marked@18`** — works fine in practice; the upstream peer range is just stale.
-2. **`deprecated prebuild-install@7.1.3`** — a transitive native-binary helper used by another dependency. The pinned `wreq-js@3.0.0` package bundles its seven supported platform addons directly; this warning does not diagnose the web-cookie transport.
+2. **`deprecated prebuild-install@7.1.3`** — the native-binary fetch helper. Only relevant later if a web-cookie provider reports a missing `tls-client-node` native binary (a separate issue, not caused by this warning).

 **No action needed** — the warnings cannot be fully silenced without forking upstream packages.

@@ -148,10 +148,9 @@ desktop app, for example:
 - `resources/app/.build/next/node_modules/playwright-<hash>/lib/…/agentParser.js` and
  `workerProcessEntry.js` — [Playwright](https://playwright.dev), the browser-automation
  library used for in-app provider login and browser-backed chat.
- `resources/app/.build/next/node_modules/wreq-js-<hash>/rust/wreq-js.win32-x64-msvc.node`
-  — the declared MIT-licensed native addon from pinned `wreq-js@3.0.0`, used for
-  browser-fingerprinted HTTP on some web providers. Its expected SHA-256 is recorded in
-  `config/release/wreq-js-native-manifest.json`.
+- `resources/app/.build/next/node_modules/tls-client-node-<hash>/bin/tls-client-windows-64-<ver>.dll`
+  — the native binary from `tls-client-node`, used for Cloudflare-tolerant HTTP on some web
+  providers.

 **Why it fires:** the Windows installer is **not yet code-signed**, so an unsigned NSIS
 installer has zero reputation and behavioral heuristics run at maximum aggression. Combined
--- a/docs/i18n/ar/llm.txt
+++ b/docs/i18n/ar/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/az/llm.txt
+++ b/docs/i18n/az/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/bg/llm.txt
+++ b/docs/i18n/bg/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/bn/llm.txt
+++ b/docs/i18n/bn/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/cs/llm.txt
+++ b/docs/i18n/cs/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/da/llm.txt
+++ b/docs/i18n/da/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/de/llm.txt
+++ b/docs/i18n/de/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/es/llm.txt
+++ b/docs/i18n/es/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/fa/llm.txt
+++ b/docs/i18n/fa/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/fi/llm.txt
+++ b/docs/i18n/fi/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/fr/llm.txt
+++ b/docs/i18n/fr/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/gu/llm.txt
+++ b/docs/i18n/gu/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/he/llm.txt
+++ b/docs/i18n/he/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/hi/llm.txt
+++ b/docs/i18n/hi/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/hu/llm.txt
+++ b/docs/i18n/hu/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/id/llm.txt
+++ b/docs/i18n/id/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/in/llm.txt
+++ b/docs/i18n/in/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/it/llm.txt
+++ b/docs/i18n/it/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/ja/llm.txt
+++ b/docs/i18n/ja/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/ko/llm.txt
+++ b/docs/i18n/ko/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/mr/llm.txt
+++ b/docs/i18n/mr/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/ms/llm.txt
+++ b/docs/i18n/ms/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/nl/llm.txt
+++ b/docs/i18n/nl/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/no/llm.txt
+++ b/docs/i18n/no/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/phi/llm.txt
+++ b/docs/i18n/phi/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/pl/llm.txt
+++ b/docs/i18n/pl/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/pt-BR/llm.txt
+++ b/docs/i18n/pt-BR/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/pt/llm.txt
+++ b/docs/i18n/pt/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/ro/llm.txt
+++ b/docs/i18n/ro/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/ru/llm.txt
+++ b/docs/i18n/ru/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/sk/llm.txt
+++ b/docs/i18n/sk/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/sv/llm.txt
+++ b/docs/i18n/sv/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/sw/llm.txt
+++ b/docs/i18n/sw/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/ta/llm.txt
+++ b/docs/i18n/ta/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/te/llm.txt
+++ b/docs/i18n/te/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/th/llm.txt
+++ b/docs/i18n/th/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/tr/llm.txt
+++ b/docs/i18n/tr/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/uk-UA/llm.txt
+++ b/docs/i18n/uk-UA/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/ur/llm.txt
+++ b/docs/i18n/ur/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/vi/llm.txt
+++ b/docs/i18n/vi/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/zh-CN/llm.txt
+++ b/docs/i18n/zh-CN/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/i18n/zh-TW/llm.txt
+++ b/docs/i18n/zh-TW/llm.txt
@@ -227,7 +227,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -267,7 +267,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -286,7 +286,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -445,7 +445,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -480,7 +480,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/docs/reference/ENVIRONMENT.md
+++ b/docs/reference/ENVIRONMENT.md
@@ -764,18 +764,15 @@ REQUEST_TIMEOUT_MS (global override)
 | `OMNIROUTE_PROVIDER_PROBE_TIMEOUT_MS`     | `8000`               | Timeout (ms) for the `validationRead` and `modelsProbe` presets in `src/shared/network/safeOutboundFetch.ts`. Raise for slow endpoints (Cerebras, Cloudflare AI, Groq) to prevent flapping between active/error in the dashboard. Falls back to 8000ms for invalid (<1000) or non-numeric values. |
 | `OMNIROUTE_RELAY_FETCH_TIMEOUT_MS`        | `25000`               | Relay-specific fetch timeout in `open-sse/utils/proxyFetch.ts` (#9158). A hung relay must fail before the client/agent timeout (~30s) so callers see a relay-specific failure instead of a generic upstream timeout. Capped at `29000` so it always fires first. |
 | `OMNIROUTE_RETRY_BACKOFF_MS`              | `10`                  | Shared retry backoff for the direct/relay/proxy retry-once paths in `open-sse/utils/proxyFetch.ts` (#9158). `0` = retry immediately.                            |
-| `OMNIROUTE_CHATGPT_TLS_TIMEOUT_MS`        | `60000`              | Native wreq-js request timeout (`chatgptTlsClient.ts`).                                                                                                          |
-| `OMNIROUTE_CHATGPT_TLS_GRACE_MS`          | `10000`              | JS-side hard-deadline grace added on top of the native timeout.                                                                                                  |
-| `OMNIROUTE_CHATGPT_STREAM_FIRST_BYTE_TIMEOUT_MS` | `30000` (30s) | Max wait for the first streamed byte before ChatGPT switches to a buffered response; the hard request deadline remains active.                                   |
-| `OMNIROUTE_CLAUDE_TLS_TIMEOUT_MS`         | `60000`              | Native wreq-js request timeout (`claudeTlsClient.ts`).                                                                                                           |
-| `OMNIROUTE_CLAUDE_TLS_GRACE_MS`           | `10000`              | JS-side hard-deadline grace added on top of the native timeout.                                                                                                  |
-| `OMNIROUTE_PPLX_TLS_TIMEOUT_MS`           | `30000`              | Native wreq-js request timeout (`perplexityTlsClient.ts`).                                                                                                       |
-| `OMNIROUTE_PPLX_TLS_GRACE_MS`             | `10000`              | JS-side hard-deadline grace added on top of the native timeout.                                                                                                  |
+| `OMNIROUTE_CLAUDE_TLS_TIMEOUT_MS`         | `60000`              | Wire-level timeout for the bogdanfinn/tls-client koffi binding (`claudeTlsClient.ts`).                                                                          |
+| `OMNIROUTE_CLAUDE_TLS_GRACE_MS`           | `10000`              | JS-side grace added on top of the wire timeout when the native binding is wedged.                                                                               |
+| `OMNIROUTE_PPLX_TLS_TIMEOUT_MS`           | `30000`              | Wire-level timeout for the bogdanfinn/tls-client koffi binding (`perplexityTlsClient.ts`).                                                                      |
+| `OMNIROUTE_PPLX_TLS_GRACE_MS`             | `10000`              | JS-side grace added on top of the wire timeout when the native binding is wedged.                                                                               |
 | `OMNIROUTE_PPLX_SEARCH_HINT`              | `0` (off)            | Appends "You have built-in web search. Answer questions directly using search results." to the caller's system message (`perplexity-web/protocol.ts`). Off by default — Perplexity searches anyway, and the sentence leaks into replies as meta-commentary for coding clients. Set `1`/`true`/`yes`/`on` to restore. |
-| `OMNIROUTE_GROK_TLS_TIMEOUT_MS`           | `60000`              | Native wreq-js request timeout (`grokTlsClient.ts`).                                                                                                             |
-| `OMNIROUTE_GROK_TLS_GRACE_MS`             | `10000`              | JS-side hard-deadline grace added on top of the native timeout.                                                                                                  |
-| `OMNIROUTE_NOTION_TLS_TIMEOUT_MS`         | `30000`              | Native wreq-js request timeout (`notionTlsClient.ts`); `notion-web` raises it per request to `180000` for long generations.                                      |
-| `OMNIROUTE_NOTION_TLS_GRACE_MS`           | `10000`              | JS-side hard-deadline grace added on top of the native timeout.                                                                                                  |
+| `OMNIROUTE_GROK_TLS_TIMEOUT_MS`           | `60000`              | Wire-level timeout for the bogdanfinn/tls-client koffi binding (`grokTlsClient.ts`).                                                                            |
+| `OMNIROUTE_GROK_TLS_GRACE_MS`             | `10000`              | JS-side grace added on top of the wire timeout when the native binding is wedged.                                                                               |
+| `OMNIROUTE_NOTION_TLS_TIMEOUT_MS`         | `30000`              | Wire-level timeout for the bogdanfinn/tls-client koffi binding (`notionTlsClient.ts`); the `notion-web` executor raises it per-request to `180000` for long generations. |
+| `OMNIROUTE_NOTION_TLS_GRACE_MS`           | `10000`              | JS-side grace added on top of the wire timeout when the native binding is wedged.                                                                               |
 | `OMNIROUTE_BROWSER_POOL`                  | `on`                 | Shared Playwright browser pool for browser-backed web-cookie chat (`browserPool.ts`); set `off` to disable.                                                     |
 | `WEB_COOKIE_USE_BROWSER`                  | `0`                  | Opt a web-cookie chat request into the browser-backed path (`browserBackedChat.ts`); `1` to enable.                                                             |
 | `KIMI_WEB_BASE_URL`                       | `https://www.kimi.ai` | Base URL for the Kimi Web (international kimi.ai Connect-RPC) executor (`kimi-web.ts`); override only for mirror/proxy endpoints.                               |
--- a/docs/reference/FREE_TIERS.md
+++ b/docs/reference/FREE_TIERS.md
@@ -54,10 +54,10 @@ A 50-agent web-research pass (official docs + last-7-days news, adversarially ve
 OmniRoute answers "is it free?" through two regimes that intentionally read
 different sources:

-| Regime | Source of truth | Surfaces |
-|---|---|---|
-| **Counting / displaying** | Resolved catalog — the shipped baseline overlaid by the Radar feed (`getRadarCatalog`) | Free-tier totals, budget card, dashboards |
-| **Deciding** | Shipped catalog only (`FREE_MODEL_BUDGETS` in `open-sse/config/freeModelCatalog.data.ts`) plus the local heuristics (`:free` suffix, zero pricing, `grantsFreeAccess`) | Every consumer of `src/shared/utils/freeModels.ts`: model import, `auto/*` routing, `GET /v1/models`, and the browser previews |
+| Regime                    | Source of truth                                                                                                                                                        | Surfaces                                                                                                                       |
+| ------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------ |
+| **Counting / displaying** | Resolved catalog — the shipped baseline overlaid by the Radar feed (`getRadarCatalog`)                                                                                 | Free-tier totals, budget card, dashboards                                                                                      |
+| **Deciding**              | Shipped catalog only (`FREE_MODEL_BUDGETS` in `open-sse/config/freeModelCatalog.data.ts`) plus the local heuristics (`:free` suffix, zero pricing, `grantsFreeAccess`) | Every consumer of `src/shared/utils/freeModels.ts`: model import, `auto/*` routing, `GET /v1/models`, and the browser previews |

 Counting can improve whenever a feed is available. Deciding stays on the
 release artifact, so the answer is identical in the browser and on the server,
@@ -68,7 +68,8 @@ purpose.

 ## Methodology & caveats

- Numbers are **upper-bound estimates** from each provider's documented free-tier limits as of **2026-06-17**, gathered by web research (confidence tagged per row). Free tiers change constantly — re-verify before relying on a figure.
+- Numbers are **upper-bound estimates** from each provider's documented free-tier limits as of **2026-06-17**, gathered by web research. Free tiers change constantly — re-verify before relying on a figure.
+- **What an entry actually vouches for.** No entry carries a per-row confidence rating, and the API serves none — treat every figure above as an estimate of the same, unstated quality. Two facts are different, because they are curated by hand rather than inferred: 7 entries carry an independently documented hard stop, and 13 entries carry a prompt-training disclosure. `hardStopGuaranteed` is set only when the provider's own terms say that exceeding the free allowance refuses the request rather than silently starting to bill you, with the source in a comment next to the entry; it is never defaulted to `true`, and an entry nobody has verified stays unset. So a missing hard-stop flag means "not established", not "known to bill you".
 - `estMonthlyFreeTokens` = recurring monthly tokens only. **One-time signup credits do not recur** and count as 0. Discontinued tiers are also 0.
 - Daily token cap → `monthly = daily × 30`. Only RPD documented → `RPD × ~800 output tokens × 30`. Only RPM/TPM (no daily cap) → **uncapped** (see below).
 - **Permanently free, but no published token cap** (`siliconflow`, `glm-cn`, `tencent`, `baidu`, `kilo-gateway`, `opencode-zen`): these are real recurring free access, rate/concurrency-limited. We classify them `recurring-uncapped` and **never sum them** — multiplying `RPM × 24/7 × 30d` would produce a fantasy ceiling (the inflation we reject). They are listed so you know they exist.
--- a/docs/routing/AUTO-COMBO.md
+++ b/docs/routing/AUTO-COMBO.md
@@ -184,54 +184,64 @@ See [#7992](https://github.com/diegosouzapw/OmniRoute/issues/7992) and [#7111](h

 ## How It Works (Persisted Auto-Combos)

-The Auto-Combo Engine dynamically selects the best provider/model for each request using a **15-factor scoring function** (defined in `open-sse/services/autoCombo/scoring.ts` → `DEFAULT_WEIGHTS`). The default weights sum to `1.0`; custom weights are renormalized by `normalizeScoringWeights()`. Two of the fifteen — `cacheAffinity` and `resetWindowAffinity` — carry a default weight of `0`: they are still computed for every candidate, and `cacheAffinity` gates prompt-cache deduplication outside the score, so they are declared factors that simply do not vote by default.
+The Auto-Combo Engine dynamically selects the best provider/model for each request using a **16-factor scoring function** (defined in `open-sse/services/autoCombo/scoring.ts` → `DEFAULT_WEIGHTS`). The default weights sum to `1.0`; custom weights are renormalized by `normalizeScoringWeights()`. Three of the sixteen — `cacheAffinity`, `resetWindowAffinity` and `reliability` — carry a default weight of `0`: they are still computed for every candidate, and `cacheAffinity` gates prompt-cache deduplication outside the score, so they are declared factors that simply do not vote by default.

-![Auto-Combo 15-factor scoring](../diagrams/exported/auto-combo-scoring.svg)
+![Auto-Combo 16-factor scoring](../diagrams/exported/auto-combo-scoring.svg)

-> Source: [diagrams/auto-combo-scoring.mmd](../diagrams/auto-combo-scoring.mmd) (regenerate via `npm run docs:render-diagrams`). The filename is historical; the source and rendered diagram show all 15 factors declared in `DEFAULT_WEIGHTS`.
+> Source: [diagrams/auto-combo-scoring.mmd](../diagrams/auto-combo-scoring.mmd) (regenerate via `npm run docs:render-diagrams`). The filename is historical; the source and rendered diagram show all 16 factors declared in `DEFAULT_WEIGHTS`.

-| Factor                | Default Weight | Description                                                                                                                                                                                 |
-| :-------------------- | :------------- | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
-| `quota`               | 0.1429         | Remaining quota / rate-limit headroom [0..1]                                                                                                                                                |
-| `health`              | 0.1605         | Health score from circuit breaker (CLOSED=1.0, HALF_OPEN=0.5, OPEN=0.0)                                                                                                                     |
-| `costInv`             | 0.1429         | Inverse **blended** cost (60% input + 40% output token price, normalized) — cheaper = higher score                                                                                          |
-| `latencyInv`          | 0.1143         | Inverse p95 latency normalized to pool — faster = higher score                                                                                                                              |
-| `taskFit`             | 0.0762         | Task-type fitness (coding, review, planning, analysis, debugging, docs)                                                                                                                     |
-| `stability`           | 0.0476         | Variance-based stability (low latency stdDev / error rate)                                                                                                                                  |
-| `tierPriority`        | 0.0476         | Account-tier priority — Ultra=1.0, Pro=0.67, Standard=0.33, Free=0.0                                                                                                                        |
-| `tierAffinity`        | 0.0476         | Affinity between the candidate's tier and the manifest-recommended tier                                                                                                                     |
-| `specificityMatch`    | 0.0476         | Match between request specificity (manifest hint) and model tier                                                                                                                            |
-| `contextAffinity`     | 0.0476         | Affinity between the request's context-window need and the model's context window                                                                                                           |
-| `sessionAvailability` | 0.0476         | OAuth session availability of the candidate connection for this session (`getOAuthSessionAvailability()`; non-OAuth connections score 1.0)                                                  |
-| `connectionDensity`   | 0.0476         | Spreads load across connections of the same provider (anti-concentration)                                                                                                                   |
-| `cacheAffinity`       | 0.00           | Rendezvous-hash affinity toward the connection likeliest to already hold this request's prompt-cache prefix (`open-sse/services/combo/promptCacheAffinity.ts`); disabled by default (#8008) |
-| `resetWindowAffinity` | 0.00           | Bias toward connections whose quota reset window is favorable (disabled by default)                                                                                                         |
-| `quality`             | 0.03           | Feedback-driven output-quality signal from the routing-event quality tracker; candidates without observations receive a neutral 0.5                                                         |
+| Factor                | Default Weight | Description                                                                                                                                                                                    |
+| :-------------------- | :------------- | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| `quota`               | 0.1429         | Remaining quota / rate-limit headroom [0..1]                                                                                                                                                   |
+| `health`              | 0.1605         | Health score from circuit breaker (CLOSED=1.0, HALF_OPEN=0.5, OPEN=0.0)                                                                                                                        |
+| `costInv`             | 0.1429         | Inverse **blended** cost (60% input + 40% output token price, normalized) — cheaper = higher score                                                                                             |
+| `latencyInv`          | 0.1143         | Inverse p95 latency normalized to pool — faster = higher score                                                                                                                                 |
+| `taskFit`             | 0.0762         | Task-type fitness (coding, review, planning, analysis, debugging, docs)                                                                                                                        |
+| `stability`           | 0.0476         | Variance-based stability from latency standard deviation — a candidate whose response time swings scores lower                                                                                 |
+| `tierPriority`        | 0.0476         | Account-tier priority — Ultra=1.0, Pro=0.67, Standard=0.33, Free=0.0                                                                                                                           |
+| `tierAffinity`        | 0.0476         | Affinity between the candidate's tier and the manifest-recommended tier                                                                                                                        |
+| `specificityMatch`    | 0.0476         | Match between request specificity (manifest hint) and model tier                                                                                                                               |
+| `contextAffinity`     | 0.0476         | Affinity between the request's context-window need and the model's context window                                                                                                              |
+| `sessionAvailability` | 0.0476         | OAuth session availability of the candidate connection for this session (`getOAuthSessionAvailability()`; non-OAuth connections score 1.0)                                                     |
+| `connectionDensity`   | 0.0476         | Spreads load across connections of the same provider (anti-concentration)                                                                                                                      |
+| `cacheAffinity`       | 0.00           | Rendezvous-hash affinity toward the connection likeliest to already hold this request's prompt-cache prefix (`open-sse/services/combo/promptCacheAffinity.ts`); disabled by default (#8008)    |
+| `resetWindowAffinity` | 0.00           | Bias toward connections whose quota reset window is favorable (disabled by default)                                                                                                            |
+| `quality`             | 0.03           | Feedback-driven output-quality signal from the routing-event quality tracker; candidates without observations receive a neutral 0.5                                                            |
+| `reliability`         | 0.00           | Observed success share, `1 - failureRate`, from 24h of usage history behind a ten-sample floor (real-time metrics otherwise); candidates with no observations read as 1.0. Disabled by default |

-**Sum:** `0.1429 + 0.1605 + 0.1429 + 0.1143 + 0.0762 + (7 × 0.0476) + 0.00 + 0.00 + 0.03 = 1.0` as declared in `DEFAULT_WEIGHTS`; user-configured weights are renormalized into a distribution by `normalizeScoringWeights()` before scoring.
+**Sum:** `0.1429 + 0.1605 + 0.1429 + 0.1143 + 0.0762 + (7 × 0.0476) + 0.00 + 0.00 + 0.03 + 0.00 = 1.0` as declared in `DEFAULT_WEIGHTS`; user-configured weights are renormalized into a distribution by `normalizeScoringWeights()` before scoring.

 ## Mode Packs

-Six pre-defined weight profiles in `open-sse/services/autoCombo/modePacks.ts` — `ship-fast`, `cost-saver`, `quality-first`, `offline-friendly`, `reliability-first` and `chaos-mode` (fault-injection). Each pack overrides the default weights to bias selection toward a specific goal; the seed weights below are renormalized to sum 1.0 at runtime together with the session/context factors every pack also sets. The table shows the four original packs — see `modePacks.ts` for `reliability-first` and `chaos-mode`.
+6 pre-defined weight profiles in `open-sse/services/autoCombo/modePacks.ts`. Each pack replaces the default weights outright to bias selection toward one goal. Every pack already sums to `1.0` (`0.9999` as printed at four decimals), so `normalizeScoringWeights()` has nothing meaningful to correct when a pack is active — the values below are, to rounding, the ones the scorer applies.

-| Factor       | ship-fast | cost-saver | quality-first | offline-friendly |
-| :----------- | :-------- | :--------- | :------------ | :--------------- |
-| quota        | 0.14      | 0.14       | 0.10          | **0.37**         |
-| health       | 0.28      | 0.19       | 0.18          | 0.28             |
-| costInv      | 0.05      | **0.37**   | 0.05          | 0.10             |
-| latencyInv   | **0.32**  | 0.05       | 0.05          | 0.05             |
-| taskFit      | 0.10      | 0.10       | **0.37**      | 0.00             |
-| stability    | 0.00      | 0.05       | 0.15          | 0.10             |
-| tierPriority | 0.05      | 0.05       | 0.05          | 0.05             |
+| Factor                | ship-fast  | cost-saver | quality-first | offline-friendly | reliability-first | chaos-mode |
+| :-------------------- | :--------- | :--------- | :------------ | :--------------- | :---------------- | :--------- |
+| `quota`               | 0.1333     | 0.1333     | 0.0952        | **0.3524**       | 0.1333            | 0.0476     |
+| `health`              | 0.2667     | 0.1810     | 0.1714        | 0.2667           | **0.3524**        | **0.4000** |
+| `costInv`             | 0.0476     | **0.3524** | 0.0476        | 0.0952           | 0.0381            | 0.0190     |
+| `latencyInv`          | **0.3048** | 0.0476     | 0.0476        | 0.0476           | 0.0476            | 0.0286     |
+| `taskFit`             | 0.0952     | 0.0952     | **0.3524**    | 0.0000           | 0.0952            | 0.1905     |
+| `stability`           | 0.0000     | 0.0476     | 0.1429        | 0.0952           | 0.1905            | 0.1714     |
+| `tierPriority`        | 0.0476     | 0.0476     | 0.0476        | 0.0476           | 0.0476            | 0.0190     |
+| `tierAffinity`        | 0.0000     | 0.0000     | 0.0000        | 0.0000           | 0.0000            | 0.0000     |
+| `specificityMatch`    | 0.0000     | 0.0000     | 0.0000        | 0.0000           | 0.0000            | 0.0000     |
+| `contextAffinity`     | 0.0095     | 0.0000     | 0.0000        | 0.0000           | 0.0000            | 0.0286     |
+| `sessionAvailability` | 0.0476     | 0.0476     | 0.0476        | 0.0476           | 0.0476            | 0.0476     |
+| `resetWindowAffinity` | 0.0000     | 0.0000     | 0.0000        | 0.0000           | 0.0000            | 0.0000     |
+| `connectionDensity`   | 0.0476     | 0.0476     | 0.0476        | 0.0476           | 0.0476            | 0.0476     |

 Notes:

- `tierAffinity` and `specificityMatch` are explicitly set to `0` in every mode pack.
+- **No pack sets `quality`, and a pack replaces the weight map wholesale** (`weights = pack`, not a merge). `quality` carries `0.03` in `DEFAULT_WEIGHTS`, but under any mode pack it normalizes to `0` — selecting a pack silences the observed-quality signal completely. If you want quality feedback to influence routing, leave `modePack` unset and tune the weights directly. (`cacheAffinity` is also unset by every pack, but it defaults to `0` anyway, so nothing changes there.)
+- `tierAffinity`, `specificityMatch` and `resetWindowAffinity` are explicitly `0` in every pack.
 - Each pack's emphasis at a glance:
-  - **ship-fast** → latencyInv 0.32 + health 0.28 (low-latency, healthy connections)
-  - **cost-saver** → costInv 0.37 (cheapest tokens win)
-  - **quality-first** → taskFit 0.37 + stability 0.15 (best model for the task, consistent)
-  - **offline-friendly** → quota 0.37 + health 0.28 (max headroom regardless of speed/cost)
+  - **ship-fast** → latencyInv 0.3048 + health 0.2667 (low-latency, healthy connections)
+  - **cost-saver** → costInv 0.3524 (cheapest tokens win)
+  - **quality-first** → taskFit 0.3524 + stability 0.1429 (best model for the task, consistent)
+  - **offline-friendly** → quota 0.3524 + health 0.2667 (max headroom regardless of speed/cost)
+  - **reliability-first** → health 0.3524 + stability 0.1905 (fewest surprises)
+  - **chaos-mode** → health 0.4000 + taskFit 0.1905 (fault-injection profile)

 ### Per-Request Controls (headers) — #6023 / #6024 / #6025 / #3470

@@ -280,7 +290,7 @@ OmniRoute's combo engine supports **19 routing strategies** (declared in `src/sh
 | `reset-window`      | Prefer targets whose quota window resets soonest                                                                                                                                          |
 | `headroom`          | Pick the target with the most remaining quota headroom                                                                                                                                    |
 | `strict-random`     | Random without deduplication of repeats                                                                                                                                                   |
-| `auto`              | Use Auto Combo scoring (15-factor) — **recommended**                                                                                                                                      |
+| `auto`              | Use Auto Combo scoring (16-factor) — **recommended**                                                                                                                                      |
 | `lkgp`              | Last-Known-Good Path (pins to the last successful provider, then falls back to rules)                                                                                                     |
 | `context-optimized` | Pick target with best fit for current context size                                                                                                                                        |
 | `cache-optimized`   | Reorder targets by prompt-cache affinity — the connection likeliest to already hold this request's cached prefix is tried first (`open-sse/services/combo/promptCacheAffinity.ts`, #8008) |
@@ -389,7 +399,7 @@ The Auto Combo engine doesn't require pre-defined combos. Instead, `open-sse/ser
 3. Cross-references with `getProviderRegistry()` for model availability + pricing
 4. For each tuple `(provider, model, connection)`, builds a `VirtualAutoComboCandidate`
 5. Picks `connection.defaultModel` (or the registry's first model) as the dispatch target
-6. Scores each candidate using the 15-factor `scorePool()` and the variant's weight pack
+6. Scores each candidate using the 16-factor `scorePool()` and the variant's weight pack
 7. Returns the resulting in-memory `AutoComboConfig` for `handleComboChat()` — never persisted to DB

 This means **adding a new provider with `auto/*` enabled automatically expands the candidate pool** — no manual combo editing needed. The virtual combo is rebuilt per request, so newly-added or newly-healthy connections are picked up immediately.
@@ -450,7 +460,7 @@ Each strategy picks one provider from the candidate pool, given a `RoutingContex
 (task type, tool/vision hints, token estimate, optional SLA policy, optional
 last-known-good provider).

-#### 1. `rules` (default) — 15-factor weighted scoring
+#### 1. `rules` (default) — 16-factor weighted scoring

 Wraps the existing scoring engine. Filters out `OPEN` circuit-breaker
 candidates, then runs `scorePool()` with the current task type and `getTaskFitness()`,
@@ -459,7 +469,7 @@ picking the top-scoring provider.
 ```ts
 class RulesStrategyImpl implements RouterStrategy {
  readonly name = "rules";
-  readonly description = "15-factor weighted scoring (see DEFAULT_WEIGHTS)";
+  readonly description = "16-factor weighted scoring (see DEFAULT_WEIGHTS)";

  select(pool, context) {
    const eligible = pool.filter((c) => c.circuitBreakerState !== "OPEN");
@@ -700,7 +710,7 @@ Including the bare `auto` (default) plus the 6 `AutoVariant` values declared in

 ## How tiers fit Auto-Combo

-The 15-factor scoring function (`open-sse/services/autoCombo/scoring.ts`) treats tier
+The 16-factor scoring function (`open-sse/services/autoCombo/scoring.ts`) treats tier
 membership as two signals: `tierPriority` (0.0476) and `tierAffinity` (0.0476). See the
 canonical [scoring factor table](#how-it-works-persisted-auto-combos) above for the full
 `DEFAULT_WEIGHTS` set — the per-pack overrides (ship-fast/cost-saver/quality-first/
@@ -753,15 +763,15 @@ intentionally excluded from CI because they require live credentials and VPS acc

 ## Files

-| File                                                      | Purpose                                                                    |
-| :-------------------------------------------------------- | :------------------------------------------------------------------------- |
-| `open-sse/services/autoCombo/scoring.ts`                  | 15-factor scoring function, `DEFAULT_WEIGHTS`, pool norm                   |
-| `open-sse/services/autoCombo/taskFitness.ts`              | Model × task fitness lookup                                                |
-| `open-sse/services/autoCombo/engine.ts`                   | Selection logic, bandit, budget cap                                        |
-| `open-sse/services/autoCombo/selfHealing.ts`              | Exclusion, probes, incident mode                                           |
-| `open-sse/services/autoCombo/modePacks.ts`                | 4 weight profiles (ship-fast, cost-saver, quality-first, offline-friendly) |
-| `open-sse/services/autoCombo/autoPrefix.ts`               | `auto/` prefix parser + 6 variants                                         |
-| `open-sse/services/autoCombo/virtualFactory.ts`           | Builds in-memory `AutoComboConfig` from live connections                   |
-| `open-sse/services/autoCombo/providerRegistryAccessor.ts` | Test hook for mocking provider registry                                    |
-| `src/shared/constants/routingStrategies.ts`               | `ROUTING_STRATEGY_VALUES` (19 strategies)                                  |
-| `src/sse/handlers/chat.ts`                                | Integration: auto-prefix short-circuit                                     |
+| File                                                      | Purpose                                                                                                   |
+| :-------------------------------------------------------- | :-------------------------------------------------------------------------------------------------------- |
+| `open-sse/services/autoCombo/scoring.ts`                  | 16-factor scoring function, `DEFAULT_WEIGHTS`, pool norm                                                  |
+| `open-sse/services/autoCombo/taskFitness.ts`              | Model × task fitness lookup                                                                               |
+| `open-sse/services/autoCombo/engine.ts`                   | Selection logic, bandit, budget cap                                                                       |
+| `open-sse/services/autoCombo/selfHealing.ts`              | Exclusion, probes, incident mode                                                                          |
+| `open-sse/services/autoCombo/modePacks.ts`                | 6 weight profiles (ship-fast, cost-saver, quality-first, offline-friendly, reliability-first, chaos-mode) |
+| `open-sse/services/autoCombo/autoPrefix.ts`               | `auto/` prefix parser + 6 variants                                                                        |
+| `open-sse/services/autoCombo/virtualFactory.ts`           | Builds in-memory `AutoComboConfig` from live connections                                                  |
+| `open-sse/services/autoCombo/providerRegistryAccessor.ts` | Test hook for mocking provider registry                                                                   |
+| `src/shared/constants/routingStrategies.ts`               | `ROUTING_STRATEGY_VALUES` (19 strategies)                                                                 |
+| `src/sse/handlers/chat.ts`                                | Integration: auto-prefix short-circuit                                                                    |
--- a/docs/routing/STRICT_ZERO_COST.md
+++ b/docs/routing/STRICT_ZERO_COST.md
@@ -116,13 +116,37 @@ no waiting out the TTL after a 402/403/quota-exhausted response.
 `freeAccessPolicy`: a candidate can be economically `SAFE` and still excluded here for
 contractual reasons, or left in when this guard is off even with `freeAccessPolicy: "strict"` on.

-## What passes today
+## Seeing what the guard excludes

-Run `npx tsx scripts/ad-hoc/dry-run-strict-zero-cost.ts` against a live instance's
-`GET /v1/auto-combo/{channel}/candidates` output for a real before/after — the script reads each
-candidate's real `connectionId`, so it also proves the connection-safety fix live, not just in
-unit tests. Keyless candidates must arrive with the synthetic no-auth `connectionId`, never a
-real connection. The current built-in keyless auto path is OpenCode Free; exact candidate counts
+`GET /v1/auto-combo/{channel}/candidates` lists every candidate, including the ones this guard
+would keep out of dispatch, and each carries `freeAccessExclusion` — `null` when the guard is
+satisfied, otherwise the reason. The listing reports; it never enforces. Turning the policy off
+leaves the field `null` everywhere and costs nothing.
+
+| `freeAccessExclusion`  | What it means                                                                                                 | What to do about it                                                                                                                 |
+| :--------------------- | :------------------------------------------------------------------------------------------------------------ | :---------------------------------------------------------------------------------------------------------------------------------- |
+| `not-in-catalog`       | The provider/model pair is absent from `FREE_MODEL_BUDGETS`.                                                  | Add a curated entry, or accept that new pairs start excluded — that is the design.                                                  |
+| `regime-not-free`      | Catalogued, but its `freeType` is not one that grants free access (a discontinued tier, for instance).        | Nothing to fix. The model costs money.                                                                                              |
+| `no-hard-stop`         | Free regime, but `hardStopGuaranteed` is not `true`, so exceeding the allowance might silently start billing. | Verify the provider's terms and set the flag with the source in a comment — never to grow the catalog.                              |
+| `contradictory-noauth` | A no-auth candidate whose catalog entry is not `keyless`. Fail-closed on inconsistent metadata.               | Fix the catalog entry; the two facts disagree.                                                                                      |
+| `exhausted`            | A fresh reading says the allowance is used up.                                                                | Wait for the reset. This one resolves itself.                                                                                       |
+| `state-unknown`        | No quota reading, or one too old to trust.                                                                    | Go look: the provider may have no usage adapter registered, or the quota fetch is failing.                                          |
+| `no-connection`        | The candidate carries no account to check at all.                                                             | Not a quota problem: the candidate was built without a connection, so nothing was ever looked up. Check how the pool was assembled. |
+
+The last two are the pair worth separating. An exhausted allowance resets on its own; a reading
+that never arrives means the lookup itself is broken, and until now both looked identical from
+outside — the candidate simply vanished.
+
+**One gap remains, and it is deliberate.** `excludeTosAvoid` still removes candidates before the
+listing is built, so a model curated `tos: "avoid"` is absent with no reason given — the same
+invisibility this section just closed for the zero-cost guard. Closing it too means deciding what
+a ToS exclusion should report, which is a separate question from economic safety; this page names
+the gap rather than pretending it is not there.
+
+For an offline before/after, `npx tsx scripts/ad-hoc/dry-run-strict-zero-cost.ts` still works
+against a live instance's candidates output; it reads each candidate's real `connectionId`, so it
+also exercises the connection-safety path. Keyless candidates must arrive with the synthetic
+no-auth `connectionId`, never a real connection. The current built-in keyless auto path is OpenCode Free; exact candidate counts
 still depend on live model discovery and should be measured on the target deployment instead of
 copied from an older run. A `recurring-*` candidate passes only when it has both a registered
 usage adapter and `hardStopGuaranteed: true`; incomplete metadata remains fail-closed.
--- a/docs/security/STEALTH_GUIDE.md
+++ b/docs/security/STEALTH_GUIDE.md
@@ -1,13 +1,13 @@
 ---
 title: "Stealth Guide"
-version: 3.8.50
-lastUpdated: 2026-08-26
+version: 3.8.40
+lastUpdated: 2026-06-28
 ---

 # Stealth Guide

-> **Source of truth:** `open-sse/utils/tlsClient.ts`, `open-sse/services/{tlsClientBase,chatgptTlsClient,claudeTlsClient,perplexityTlsClient,grokTlsClient,notionTlsClient,lmarenaTlsClient,claudeCodeCCH,claudeCodeFingerprint,claudeCodeObfuscation,claudeCodeCompatible}.ts`, `open-sse/config/cliFingerprints.ts`, `src/mitm/`
-> **Last updated:** 2026-08-26 — v3.8.50
+> **Source of truth:** `open-sse/utils/tlsClient.ts`, `open-sse/services/{claudeCodeCCH,claudeCodeFingerprint,claudeCodeObfuscation,claudeCodeCompatible}.ts`, `open-sse/config/cliFingerprints.ts`, `src/mitm/`
+> **Last updated:** 2026-06-28 — v3.8.40
 > **Audience:** Engineers maintaining provider-specific stealth integrations.

 OmniRoute integrates with providers whose edges actively fingerprint non-official clients (TLS JA3/JA4, header ordering, JSON body shape, integrity tokens). This page documents the stealth surfaces OmniRoute exposes and where they are implemented.
@@ -29,41 +29,6 @@ Lazy-loaded `wreq-js` session that impersonates **Chrome 124 on macOS**. Used as
 - Timeout: `TLS_CLIENT_TIMEOUT_MS` (inherits from `FETCH_TIMEOUT_MS`, default 600000)
 - `wreq-js` Response is fetch-compatible (`headers`, `text()`, `json()`, `clone()`, `body`).

-### Web-cookie provider transport — wreq-js 3.0.0
-
-`open-sse/services/tlsClientBase.ts` is the shared transport for ChatGPT, Claude, Perplexity,
-Grok, Notion, and LMArena web sessions. Each thin provider wrapper selects a browser/OS profile;
-the base loads `wreq-js` lazily, reuses only transport-level connections keyed by
-profile + OS + resolved proxy, and gives every request an ephemeral cookie scope. It never shares a
-wreq session or cookie jar between accounts or requests.
-
-| Provider   | Profile       | Emulated OS | Stream EOF policy                |
-| ---------- | ------------- | ----------- | -------------------------------- |
-| ChatGPT    | `firefox_148` | macOS       | include `[DONE]`                 |
-| Claude     | `chrome_146`  | Linux       | include `[DONE]`                 |
-| Perplexity | `firefox_148` | macOS       | include `event: end_of_stream`   |
-| Grok       | `chrome_146`  | Linux       | exclude `[DONE]`                 |
-| Notion     | `chrome_146`  | Windows     | include `[DONE]`                 |
-| LMArena    | `chrome_146`  | Windows     | no sentinel; close on native EOF |
-
- Streaming uses the native response `ReadableStream` directly; no temp file or sidecar is created.
- Up to 256 initial bytes are inspected before exposing a stream. SSE providers buffer non-SSE
-  errors; Grok/LMArena map Cloudflare challenges to `403` and HTML interstitials to `502`.
- The native request timeout remains wrapped by an absolute JS hard deadline. A hang invalidates
-  and closes only the affected profile/OS/proxy transport before the next request recreates it.
- Proxy resolution priority is per-call `proxyUrl` → request-scoped account/dashboard context →
-  `HTTPS_PROXY`/`HTTP_PROXY`/`ALL_PROXY` (including lowercase variants). Resolution errors fail
-  closed instead of leaking a direct connection. LMArena deliberately resolves against `arena.ai`.
- `byteResponse` returns a content-typed `data:` URL without UTF-8 corruption.
- Errors are `TlsClientUnavailableError` (package/addon unavailable) and `TlsClientHangError`
-  (deadline exceeded).
-
-The profiles are supported by the pinned package, but real WAF acceptance can change independently
-of local contract tests. Validate fingerprint changes against an explicitly authorized live account
-before claiming parity with an upstream browser.
-
---
-
 ## Claude Code Stealth Bundle

 When `cliCompatMode` is on, OmniRoute reshapes outgoing Claude requests so they are indistinguishable from `claude-cli` traffic. Three modules collaborate:
--- a/llm.txt
+++ b/llm.txt
@@ -223,7 +223,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   │   ├── accountFallback.ts    # Multi-account fallback
 │   │   ├── sessionManager.ts     # Session management
 │   │   ├── wildcardRouter.ts     # Wildcard model routing
-│   │   ├── autoCombo/            # Auto-combo engine (15-factor scoring, bandit exploration)
+│   │   ├── autoCombo/            # Auto-combo engine (16-factor scoring, bandit exploration)
 │   │   ├── intentClassifier.ts   # Request intent classification
 │   │   ├── taskAwareRouter.ts    # Task-aware routing
 │   │   ├── thinkingBudget.ts     # Thinking budget management
@@ -263,7 +263,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 │   ├── architecture/             # ARCHITECTURE.md, CODEBASE_DOCUMENTATION.md, REPOSITORY_MAP.md, AUTHZ_GUIDE.md, RESILIENCE_GUIDE.md, QUALITY_GATES.md
 │   ├── reference/                # API_REFERENCE.md, PROVIDER_REFERENCE.md, CLI-TOOLS.md
 │   ├── frameworks/               # MCP-SERVER.md (110 tools), A2A-SERVER.md, SKILLS.md, MEMORY.md, CLOUD_AGENT.md, EVALS.md, WEBHOOKS.md
-│   ├── routing/                  # AUTO-COMBO.md (15-factor scoring), REASONING_REPLAY.md
+│   ├── routing/                  # AUTO-COMBO.md (16-factor scoring), REASONING_REPLAY.md
 │   ├── security/                 # GUARDRAILS.md, COMPLIANCE.md, STEALTH_GUIDE.md, PUBLIC_CREDS.md, ERROR_SANITIZATION.md
 │   ├── guides/                   # USER_GUIDE.md, TROUBLESHOOTING.md, ELECTRON_GUIDE.md, I18N.md
 │   ├── ops/                      # RELEASE_CHECKLIST.md, TUNNELS_GUIDE.md, VM deployment
@@ -282,7 +282,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 - **19 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, cache-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
 - **Context Relay strategy**: Session handoff summaries on account rotation for continuity
- **Auto-combo engine**: Self-healing routing optimization with **15-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
+- **Auto-combo engine**: Self-healing routing optimization with **16-factor scoring** (see `docs/routing/AUTO-COMBO.md`), bandit exploration, progressive cooldown
 - **Semantic caching** with cache hit/miss headers
 - **Idempotency** with configurable dedup window
 - **3-layer resilience**: Provider Circuit Breaker / Connection Cooldown / Model Lockout
@@ -441,7 +441,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 8. **ACP agents** are in `src/lib/acp/registry.ts` with detection cache. Custom agents stored via settings DB.

-9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **15-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.
+9. **Auto-combo engine** in `open-sse/services/autoCombo/` — **16-factor scoring** (weights and factors in `docs/routing/AUTO-COMBO.md`), 6 mode packs, bandit exploration, progressive cooldown.

 10. **Docker:** Dockerfile has two targets: `runner-base` and `runner-cli`. `docker-compose.yml` for dev (8 profiles), `docker-compose.prod.yml` for production (port 20130).

@@ -476,7 +476,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool

 - **352-provider catalog** with 150+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
- **15-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
+- **16-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)
 - **Cloud Agents** (Codex Cloud, Cursor, Devin, Jules), **Guardrails**, **Evals**, **Webhooks**, **Compliance** frameworks
 - **Embedded services** manager (install/start/stop bundled services from the dashboard)
--- a/next.config.mjs
+++ b/next.config.mjs
@@ -339,6 +339,9 @@ const nextConfig = {
    "keytar",
    "wreq-js",
    "zod",
+    "tls-client-node",
+    "koffi",
+    "tough-cookie",
    "@ngrok/ngrok",
    "@huggingface/transformers",
    // The ESM entry imports tiktoken_bg.wasm as a module. Turbopack can compile
--- a/open-sse/executors/grok-web.ts
+++ b/open-sse/executors/grok-web.ts
@@ -939,8 +939,8 @@ export class GrokWebExecutor extends BaseExecutor {

    // Fetch from Grok via TLS-impersonating client (#3180).
    // Grok sits behind Cloudflare Enterprise which rejects Node's native TLS
-    // fingerprint even with valid sso+sso-rw cookies. The pinned wreq-js
-    // transport sends a Chrome-like handshake instead.
+    // fingerprint even with valid sso+sso-rw cookies. We use tls-client-node
+    // to send a Chrome-like handshake instead.
    let tlsResult: TlsFetchResult;
    try {
      tlsResult = await tlsFetchGrok(GROK_CHAT_API, {
--- a/open-sse/executors/lmarena.ts
+++ b/open-sse/executors/lmarena.ts
@@ -2,8 +2,8 @@
 * LMArenaExecutor — Arena (formerly LMArena) web-session provider.
 *
 * Routes requests through arena.ai create-evaluation with session cookies.
- * Upstream sits behind Cloudflare; traffic goes through wreq-js Chrome
- * impersonation with isolated ephemeral cookies (see services/lmarenaTlsClient.ts).
+ * Upstream sits behind Cloudflare; traffic goes through tls-client-node Chrome
+ * impersonation (see services/lmarenaTlsClient.ts).
 *
 * Helpers: open-sse/executors/lmarena/{cookie,models,stream,response}.ts
 */
@@ -174,6 +174,7 @@ export class LMArenaExecutor extends BaseExecutor {
      body: JSON.stringify(transformedBody),
      signal: ctx.signal,
      stream: ctx.stream,
+      streamEofSymbol: "__OMNIROUTE_LMARENA_EOF_NEVER__",
    });

    const failed = mapFailedTlsResult({
--- a/open-sse/executors/lmarena/models.ts
+++ b/open-sse/executors/lmarena/models.ts
@@ -6,9 +6,9 @@ export const LMARENA_API_BASE = "https://arena.ai";
 export const LMARENA_STREAM_URL = `${LMARENA_API_BASE}/nextjs-api/stream/create-evaluation`;
 /**
 * Current Chrome stable UA (header surface).
- * TLS JA3/JA4 profile is separate: the provider-tested wreq-js profile is pinned
- * to chrome_146 in lmarenaTlsClient.ts while headers track the live browser string.
- * Treat that deliberate version skew as a WAF-sensitive compatibility surface.
+ * TLS JA3 profile is separate: tls-client-node tops out at chrome_146 — see
+ * LMARENA_PROFILE in lmarenaTlsClient.ts. Headers track the live browser string;
+ * fingerprint stays at the newest native profile we can actually impersonate.
 */
 export const LMARENA_USER_AGENT =
  "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/150.0.0.0 Safari/537.36";
--- a/open-sse/executors/lmarena/response.ts
+++ b/open-sse/executors/lmarena/response.ts
@@ -114,7 +114,7 @@ export function mapTlsUnavailable(
  return {
    response: errorResponse(
      502,
-      `Arena TLS impersonation unavailable: ${error.message}. Verify the wreq-js 3.0.0 native addon.`,
+      `Arena TLS impersonation unavailable: ${error.message}. Install/repair tls-client-node native binary.`,
      "upstream_error",
      "TLS_CLIENT_UNAVAILABLE"
    ),
--- a/open-sse/executors/notion-web.ts
+++ b/open-sse/executors/notion-web.ts
@@ -22,7 +22,7 @@
 * chunk — safer than assuming unverified incremental-delta semantics.
 *
 * Auth: Cookie-based (token_v2 [+ optional space_id, notion_browser_id, user_id])
- * Method: Browser-TLS impersonation via pinned wreq-js (Chrome JA3/JA4). Plain
+ * Method: Browser-TLS impersonation via tls-client-node (Chrome JA3). Plain
 * Node/undici fetch is rejected by Notion's edge with in-band
 * `temporarily-unavailable` (HTTP 200, empty assistant text) — curl/Schannel
 * and Chrome work with the same cookie + body. See services/notionTlsClient.ts.
@@ -60,7 +60,10 @@ import {
  messagesForNotionTranscript,
  type NotionAgentOptions,
 } from "../services/notionTranscriptBuilder.ts";
-import { tlsFetchNotion, TlsClientUnavailableError } from "../services/notionTlsClient.ts";
+import {
+  tlsFetchNotion,
+  TlsClientUnavailableError,
+} from "../services/notionTlsClient.ts";

 // Re-exported for unit tests that destructure `mod.<name>` on this module.
 export {
@@ -222,6 +225,7 @@ function extractUserIdFromCookie(cookie: string): string {
  return extractNotionUserIdFromCookie(cookie);
 }

+
 /**
 * Notion's undocumented inference API does not return token usage.
 * Emit a cheap char-based estimate so clients don't see a constant
@@ -232,7 +236,9 @@ export function estimateNotionUsage(
  messages: NotionMessage[] | undefined,
  content: string
 ): { prompt_tokens: number; completion_tokens: number; total_tokens: number; estimated: true } {
-  const promptText = (messages || []).map((m) => extractNotionMessageText(m?.content)).join("\n");
+  const promptText = (messages || [])
+    .map((m) => extractNotionMessageText(m?.content))
+    .join("\n");
  // ~4 chars/token (English-ish); at least 1 when there is any text.
  const prompt_tokens = promptText ? Math.max(1, Math.ceil(promptText.length / 4)) : 0;
  const completion_tokens = content ? Math.max(1, Math.ceil(content.length / 4)) : 0;
@@ -387,8 +393,9 @@ function buildNotionExecuteHeaders(opts: {
  const isCustom = Boolean(opts.agent?.workflowId);
  // Browser uses /agent/<workflowId without dashes>?wfv=chat for custom agents.
  const agentPathId = (opts.agent?.workflowId || "").replace(/-/g, "");
-  const referer =
-    isCustom && agentPathId ? `${BASE_URL}/agent/${agentPathId}?wfv=chat` : `${BASE_URL}/ai`;
+  const referer = isCustom && agentPathId
+    ? `${BASE_URL}/agent/${agentPathId}?wfv=chat`
+    : `${BASE_URL}/ai`;
  const reqHeaders: Record<string, string> = {
    "Content-Type": "application/json",
    "User-Agent": USER_AGENT,
@@ -446,8 +453,11 @@ export function resolveNotionAgentOptions(
      "agent_id",
    ]) || "";
  const pageFromPs =
-    readProviderSpecificString(ps, ["contextPageId", "context_page_id", "notionContextPageId"]) ||
-    "";
+    readProviderSpecificString(ps, [
+      "contextPageId",
+      "context_page_id",
+      "notionContextPageId",
+    ]) || "";

  const readCookie = (name: string): string => {
    const m = cookie.match(new RegExp(`(?:^|;\\s*)${name}=([^;]+)`, "i"));
@@ -467,7 +477,10 @@ export function resolveNotionAgentOptions(
      readCookie("agent_id")
  );
  const contextPageId =
-    pageFromPs || readCookie("context_page_id") || readCookie("notion_context_page_id") || "";
+    pageFromPs ||
+    readCookie("context_page_id") ||
+    readCookie("notion_context_page_id") ||
+    "";

  return {
    workflowId: workflowId || undefined,
@@ -497,7 +510,8 @@ async function sendNotionInferenceRequest(opts: {
      body: JSON.stringify(reqBody),
      signal: signal ?? undefined,
      // Inference can take a while (tool-autoload + LLM first token).
-      timeoutMs: Number.parseInt(process.env.OMNIROUTE_NOTION_TLS_TIMEOUT_MS || "", 10) || 180_000,
+      timeoutMs:
+        Number.parseInt(process.env.OMNIROUTE_NOTION_TLS_TIMEOUT_MS || "", 10) || 180_000,
    });
    status = tlsRes.status;
    rawText = tlsRes.text ?? "";
@@ -620,7 +634,8 @@ export class NotionWebExecutor extends BaseExecutor {
    const inboundHeaders =
      (input.clientHeaders as Record<string, string> | null | undefined) ??
      ((input as { headers?: Record<string, string> }).headers as
-        Record<string, string> | undefined);
+        | Record<string, string>
+        | undefined);
    const clientThreadId = readClientThreadId(requestBody, inboundHeaders ?? undefined);
    // Namespace the thread cache PER CALLER (hash of the caller's cookie) AND by custom
    // agent, so (a) two users of the same Notion space never share a cached thread
@@ -723,10 +738,7 @@ export class NotionWebExecutor extends BaseExecutor {

    // One automatic retry for transient Notion faults — same threadId, never create again
    if (isFailedAttempt(attempt) && attempt.retryable) {
-      const delayMs =
-        process.env.NODE_ENV === "test" || process.env.VITEST
-          ? 20
-          : 700 + Math.floor(Math.random() * 400);
+      const delayMs = process.env.NODE_ENV === "test" || process.env.VITEST ? 20 : 700 + Math.floor(Math.random() * 400);
      await new Promise((r) => setTimeout(r, delayMs));
      attempt = await runOnce({ createThread: false, threadId });
    }
--- a/open-sse/executors/perplexity-web.ts
+++ b/open-sse/executors/perplexity-web.ts
@@ -501,7 +501,7 @@ export class PerplexityWebExecutor extends BaseExecutor {
        if (isCloudflareChallenge(response.text)) {
          errMsg =
            "Cloudflare blocked the request — Perplexity's edge rejected this server's TLS fingerprint " +
-            "(common on VPS/datacenter IPs). Verify the wreq-js 3.0.0 native addon, " +
+            "(common on VPS/datacenter IPs). Ensure tls-client-node is installed with its native binary, " +
            "or route perplexity-web through a residential proxy.";
          log?.error?.("PPLX-WEB", "Cloudflare challenge detected — TLS bypass failed");
        } else {
--- a/open-sse/handlers/autoComboCandidates.ts
+++ b/open-sse/handlers/autoComboCandidates.ts
@@ -30,6 +30,7 @@ import { buildErrorBody } from "@omniroute/open-sse/utils/error.ts";
 import { getCircuitBreaker } from "@/shared/utils/circuitBreaker";
 import { isModelLocked } from "@omniroute/open-sse/services/accountFallback.ts";
 import { parseModel } from "@omniroute/open-sse/services/model.ts";
+import type { StrictZeroCostExclusionReason } from "@omniroute/open-sse/services/autoCombo/strictZeroCostFilter.ts";
 import { getProviderConnectionById } from "@/lib/db/providers";
 import { getExcludedConnectionIds } from "@/lib/db/autoCandidateOverrides";

@@ -51,6 +52,13 @@ export interface AutoComboCandidateView {
  breakerState: string;
  connectionCooldown: boolean;
  modelLocked: boolean;
+  /**
+   * Why STRICT_ZERO_COST would exclude this candidate from dispatch, or null
+   * when it would not — and null as well when the policy is off, which is the
+   * default. Reported, never enforced: this listing shows the candidate either
+   * way, the routing path is what acts on it.
+   */
+  freeAccessExclusion: StrictZeroCostExclusionReason | null;
 }

 export interface AutoComboCandidatesResult {
@@ -69,6 +77,7 @@ async function decorateCandidate(candidate: {
  connectionId: string;
  model: string;
  modelStr: string;
+  freeAccessExclusion?: StrictZeroCostExclusionReason | null;
 }): Promise<AutoComboCandidateView> {
  const breaker = getCircuitBreaker(candidate.provider);
  const breakerStatus = breaker.getStatus();
@@ -111,6 +120,7 @@ async function decorateCandidate(candidate: {
    breakerState: String(breakerStatus.state),
    connectionCooldown,
    modelLocked,
+    freeAccessExclusion: candidate.freeAccessExclusion ?? null,
  };
 }

@@ -160,6 +170,7 @@ export async function getAutoComboCandidates(
    connectionId: string | null;
    allowedConnectionIds?: string[];
    model: string;
+    freeAccessExclusion?: StrictZeroCostExclusionReason | null;
  }> = Array.isArray(virtualCombo?.models) ? virtualCombo.models : [];
  // Routing keeps one logical provider/model candidate, but the management API
  // remains account-oriented so operators can inspect and toggle each fallback.
@@ -178,6 +189,7 @@ export async function getAutoComboCandidates(
        connectionId: candidate.connectionId,
        model: candidate.model,
        modelStr: candidate.model,
+        freeAccessExclusion: candidate.freeAccessExclusion,
      });
      return { ...decorated, excluded: excludedConnectionIds.has(candidate.connectionId) };
    })
--- a/open-sse/handlers/chatCore/clientUsageBuffer.ts
+++ b/open-sse/handlers/chatCore/clientUsageBuffer.ts
@@ -22,6 +22,7 @@ import {
  addBufferToUsage as defaultAddBuffer,
  filterUsageForFormat as defaultFilterUsage,
  estimateUsage as defaultEstimateUsage,
+  isEmptyUsage,
  sanitizeProviderUsageForRequest,
  type UsageLike,
 } from "../../utils/usageTracking.ts";
@@ -46,35 +47,6 @@ const DEFAULT_DEPS: ClientUsageBufferDeps = {
  estimateUsage: defaultEstimateUsage,
 };

-/** True when a usage object is present but every token field is zero/absent.
- * Web/unofficial providers often emit `{prompt_tokens:0,completion_tokens:0,total_tokens:0}`
- * because the upstream has no metering. Treating that as "has usage" makes
- * `addBufferToUsage` turn zeros into a constant `USAGE_TOKEN_BUFFER` (default 2000),
- * so every request shows exactly 2000 tokens. Prefer estimating instead. */
-function isEmptyUsage(usage: unknown): boolean {
-  if (!usage || typeof usage !== "object" || Array.isArray(usage)) return true;
-  const u = usage as Record<string, unknown>;
-  const fields = [
-    "prompt_tokens",
-    "completion_tokens",
-    "total_tokens",
-    "input_tokens",
-    "output_tokens",
-    "promptTokenCount",
-    "candidatesTokenCount",
-    "totalTokenCount",
-  ];
-  let sawNumber = false;
-  for (const key of fields) {
-    const v = u[key];
-    if (typeof v !== "number" || !Number.isFinite(v)) continue;
-    sawNumber = true;
-    if (v > 0) return false;
-  }
-  // No positive counts (or no numeric fields at all) → treat as empty.
-  return true;
-}
-
 /** context_budget_* → visible-field mapping folded back in for Claude-Code-compatible
 * responses only (see module docstring above). */
 const CONTEXT_BUDGET_TO_VISIBLE_FIELD: Record<string, string> = {
--- a/open-sse/services/tests/grokTlsClient.test.ts
+++ b/open-sse/services/tests/grokTlsClient.test.ts
@@ -1,17 +1,16 @@
 /**
 * Regression tests for the proxy-leak fix in grokTlsClient.
 *
- * Bug context (#3180): tlsFetchGrok() built its native transport options
- * without a `proxyUrl` field, so every grok-web call
+ * Bug context (#3180): tlsFetchGrok() built its native tls-client-node
+ * requestOptions without a `proxyUrl` field, so every grok-web call
 * egressed with the bare host IP regardless of the dashboard proxy config
- * or HTTP_PROXY / HTTPS_PROXY env vars. Native browser transports require the
- * resolved proxy to be passed explicitly.
+ * or HTTP_PROXY / HTTPS_PROXY env vars (the koffi-loaded Go binary does not
+ * consult Go's `http.ProxyFromEnvironment`).
 *
 * These tests pin the resolution-order contract:
 *   1. Per-call `options.proxyUrl` wins.
- *   2. Request-scoped dashboard/account proxy context.
- *   3. POSIX-standard HTTPS_PROXY / HTTP_PROXY / ALL_PROXY (and lowercase variants).
- *   4. Otherwise undefined (no proxy).
+ *   2. POSIX-standard HTTPS_PROXY / HTTP_PROXY / ALL_PROXY (and lowercase variants).
+ *   3. Otherwise undefined (no proxy).
 *
 * They also pin that the resolved proxy is actually placed on the
 * requestOptions object handed to the native binding — the original bug
--- a/open-sse/services/autoCombo/routerStrategy.ts
+++ b/open-sse/services/autoCombo/routerStrategy.ts
@@ -3,7 +3,7 @@
 *
 * Inspired by ClawRouter commit 14c83c258 "refactor: extract routing into pluggable RouterStrategy system".
 * Provides a RouterStrategy interface and built-in implementations:
- *   - RulesStrategy (default): wraps the existing 15-factor scoring engine
+ *   - RulesStrategy (default): wraps the existing 16-factor scoring engine
 *   - ScoreStrategy: highest configured weighted score, with explicit exploration
 *   - CostStrategy: always picks cheapest available model
 *   - LatencyStrategy: prioritizes low p95 latency with reliability weighting
@@ -53,7 +53,7 @@ export interface RouterStrategy {
  select(pool: ProviderCandidate[], context: RoutingContext): RoutingDecision;
 }

-// ── RulesStrategy: wraps 15-factor scoring engine ───────────────────────────
+// ── RulesStrategy: wraps 16-factor scoring engine ───────────────────────────

 function toSpeedCandidate(c: ProviderCandidate): SpeedCandidate {
  return {
@@ -87,7 +87,7 @@ function toSpeedCandidate(c: ProviderCandidate): SpeedCandidate {

 class RulesStrategyImpl implements RouterStrategy {
  readonly name = "rules";
-  readonly description = "15-factor weighted scoring (see DEFAULT_WEIGHTS)";
+  readonly description = "16-factor weighted scoring (see DEFAULT_WEIGHTS)";

  select(pool: ProviderCandidate[], context: RoutingContext): RoutingDecision {
    const eligible = pool.filter((c) => c.circuitBreakerState !== "OPEN");
--- a/open-sse/services/autoCombo/scoring.ts
+++ b/open-sse/services/autoCombo/scoring.ts
@@ -29,6 +29,13 @@ export interface ScoringFactors {
   * observed events default to neutral (0.5) and are never penalized.
   */
  quality?: number;
+  /**
+   * Observed success share over the routing window: 1 - failure rate. Optional
+   * so a candidate nobody has called yet reads as 1 rather than 0 -- it has not
+   * failed anything. That differs from `quality` on purpose: a score with no
+   * observations is neutral at 0.5, a failure rate with no observations is 0.
+   */
+  reliability?: number;
 }

 export interface ScoringWeights {
@@ -48,6 +55,8 @@ export interface ScoringWeights {
  connectionDensity: number;
  /** Weight for the feedback-driven quality factor (#feedback-foundation). */
  quality?: number;
+  /** Weight for the observed failure-rate factor. 0 by default. */
+  reliability?: number;
 }

 export const DEFAULT_WEIGHTS: ScoringWeights = {
@@ -69,6 +78,12 @@ export const DEFAULT_WEIGHTS: ScoringWeights = {
  // the new quality signal (observed output quality over time) gets a real,
  // if smaller, vote. Sum remains exactly 1.0.
  quality: 0.03,
+  // Declared but silent, like `cacheAffinity` and `resetWindowAffinity`: every
+  // candidate already carries a measured failure rate (24h of usage history
+  // behind a ten-sample floor, real-time metrics otherwise) and the scorer had
+  // no way to read it. Which weight it deserves is a product call backed by
+  // measurement, so this ships at 0 and leaves the ranking exactly as it was.
+  reliability: 0,
 };

 /** Normalize independently configured UI weights into a scoring distribution. */
@@ -162,7 +177,10 @@ export function calculateScore(factors: ScoringFactors, weights: ScoringWeights)
      (weights.connectionDensity ?? 0) * factors.connectionDensity +
      // Missing quality factor → neutral 0.5: a cold candidate is neither boosted
      // (which would let optimistic initialization dominate) nor penalized.
-      (weights.quality ?? 0) * (factors.quality ?? 0.5)
+      (weights.quality ?? 0) * (factors.quality ?? 0.5) +
+      // Missing reliability factor -> neutral 1, not 0.5: a candidate with no
+      // observations has not failed anything. See the field doc on ScoringFactors.
+      (weights.reliability ?? 0) * (factors.reliability ?? 1)
  );
 }

@@ -255,6 +273,17 @@ export function computePoolMaxima(pool: ProviderCandidate[]): PoolMaxima {
  return { maxCost, maxLatency, maxStdDev };
 }

+/**
+ * Bound an observed failure rate to [0,1], treating anything missing or
+ * non-finite as 0 (nothing observed has failed). Mirrors `toBoundedRate` in
+ * `speedRanking.ts` so both consumers of the same signal agree, including on
+ * garbage input.
+ */
+function boundedRate(value: number | null | undefined): number {
+  if (typeof value !== "number" || !Number.isFinite(value) || value < 0) return 0;
+  return Math.min(1, value);
+}
+
 export function calculateFactors(
  candidate: ProviderCandidate,
  pool: ProviderCandidate[],
@@ -292,6 +321,13 @@ export function calculateFactors(
    // Feedback quality signal; neutral 0.5 when the tracker has no data yet
    // (cold providers are neither boosted nor unfairly penalized).
    quality: clamp01(candidate.quality ?? 0.5),
+    // Same formula and same precedence as `speedRanking.ts` uses for its own
+    // reliability factor: an explicit failure rate wins over the coarser error
+    // rate, and an unobserved candidate reads as fully reliable. The rate is
+    // bounded BEFORE the subtraction, exactly as `toBoundedRate` does there --
+    // `clamp01(1 - NaN)` would be 0, i.e. "fails every call", which is the
+    // opposite of what corrupt telemetry should mean.
+    reliability: clamp01(1 - boundedRate(candidate.failureRate ?? candidate.errorRate)),
  };
 }

--- a/open-sse/services/autoCombo/strictZeroCostFilter.ts
+++ b/open-sse/services/autoCombo/strictZeroCostFilter.ts
@@ -129,25 +129,52 @@ export function findBudgetEntry(
  return catalog.find((m) => m.provider === candidate.provider && m.modelId === candidate.model);
 }

-function isConnectionStateSafe(
+/** Why the guard cannot trust a candidate right now. */
+export type StrictZeroCostExclusionReason =
+  | "not-in-catalog"
+  | "regime-not-free"
+  | "no-hard-stop"
+  | "contradictory-noauth"
+  | "exhausted"
+  | "state-unknown"
+  | "no-connection";
+
+export type StrictZeroCostVerdict =
+  { outcome: "safe"; safeConnectionIds: string[] } | { outcome: StrictZeroCostExclusionReason };
+
+/**
+ * Why one connection cannot be trusted right now. Splitting "exhausted" from
+ * "state-unknown" is the whole point: an exhausted allowance resets on its own
+ * and the operator waits, while a missing or stale reading means the quota
+ * lookup itself is not working and the operator has to go fix something.
+ * Opposite actions, and until now the same silence.
+ *
+ * Freshness is checked before status, so a stale EXHAUSTED reading reports
+ * "state-unknown" rather than asserting an exhaustion nobody has confirmed
+ * lately. The exclusion verdict is identical either way -- only the reason
+ * shown to the operator differs.
+ */
+export function classifyConnectionState(
  provider: string,
  connectionId: string,
  resolveFreeAccessState: StrictZeroCostOptions["resolveFreeAccessState"],
  options: Pick<StrictZeroCostOptions, "minRemainingAllowance" | "maxStateAgeMs" | "now">
-): boolean {
+): "safe" | "exhausted" | "state-unknown" {
  const state = resolveFreeAccessState(provider, connectionId);
-  if (!state) return false; // no usage adapter for this provider, or lookup never ran/is stale
-  if (state.status !== "SAFE") return false;
+  if (!state) return "state-unknown"; // no usage adapter for this provider, or lookup never ran

  const now = (options.now ?? Date.now)();
  const checkedAtMs = Date.parse(state.checkedAt);
-  if (!Number.isFinite(checkedAtMs) || now - checkedAtMs > options.maxStateAgeMs) return false;
+  if (!Number.isFinite(checkedAtMs) || now - checkedAtMs > options.maxStateAgeMs)
+    return "state-unknown";

-  if (state.remainingFreeAllowance === null) return false;
+  if (state.status === "EXHAUSTED") return "exhausted";
+  if (state.status !== "SAFE") return "state-unknown";
+  if (state.remainingFreeAllowance === null) return "state-unknown";
  // A negative threshold would let a negative/garbage reading pass; a caller
  // that genuinely wants "any allowance greater than zero" should pass 0.
-  if (options.minRemainingAllowance < 0) return false;
-  return state.remainingFreeAllowance > options.minRemainingAllowance;
+  if (options.minRemainingAllowance < 0) return "state-unknown";
+  return state.remainingFreeAllowance > options.minRemainingAllowance ? "safe" : "exhausted";
 }

 /**
@@ -168,7 +195,28 @@ export function evaluateCandidateConnections(
  resolveFreeAccessState: StrictZeroCostOptions["resolveFreeAccessState"],
  options: Pick<StrictZeroCostOptions, "minRemainingAllowance" | "maxStateAgeMs" | "now">
 ): string[] {
-  if (!budgetEntry) return []; // not in the catalog at all → paid, or genuinely unknown
+  const verdict = classifyStrictZeroCostCandidate(
+    candidate,
+    budgetEntry,
+    resolveFreeAccessState,
+    options
+  );
+  return verdict.outcome === "safe" ? verdict.safeConnectionIds : [];
+}
+
+/**
+ * Same decision as `evaluateCandidateConnections`, but it says why instead of
+ * answering with an empty list. The read-only candidate listing needs the why;
+ * the pool filter only needs the list, so it reads this one and throws the
+ * reason away.
+ */
+export function classifyStrictZeroCostCandidate(
+  candidate: StrictZeroCostCandidate,
+  budgetEntry: FreeModelBudget | undefined,
+  resolveFreeAccessState: StrictZeroCostOptions["resolveFreeAccessState"],
+  options: Pick<StrictZeroCostOptions, "minRemainingAllowance" | "maxStateAgeMs" | "now">
+): StrictZeroCostVerdict {
+  if (!budgetEntry) return { outcome: "not-in-catalog" }; // paid, or genuinely unknown

  const isGenuineNoAuthCandidate = candidate.connectionId === SYNTHETIC_NOAUTH_CONNECTION_ID;
  if (allowsNoAuthShortcut(budgetEntry.freeType)) {
@@ -180,30 +228,47 @@ export function evaluateCandidateConnections(
    // any other freeType, and is excluded there unless hardStopGuaranteed is
    // also set for it (which the curated catalog does not do for keyless
    // entries today, so it will correctly exclude).
-    if (isGenuineNoAuthCandidate) return [SYNTHETIC_NOAUTH_CONNECTION_ID];
+    if (isGenuineNoAuthCandidate)
+      return { outcome: "safe", safeConnectionIds: [SYNTHETIC_NOAUTH_CONNECTION_ID] };
  }
-  if (!grantsFreeAccess(budgetEntry.freeType)) return [];
-  if (isGenuineNoAuthCandidate) return []; // no-auth path but a non-keyless catalog entry: contradictory metadata, fail closed
+  if (!grantsFreeAccess(budgetEntry.freeType)) return { outcome: "regime-not-free" };
+  // no-auth path but a non-keyless catalog entry: contradictory metadata, fail closed
+  if (isGenuineNoAuthCandidate) return { outcome: "contradictory-noauth" };

  // Every remaining freeType (recurring-*, one-time-initial, a keyless entry
  // reached via a real connection, and any future type this module doesn't
  // special-case) requires a documented hard stop before any live check even
  // runs — no point burning a quota lookup on a connection we could never
  // trust regardless of its answer.
-  if (budgetEntry.hardStopGuaranteed !== true) return [];
+  if (budgetEntry.hardStopGuaranteed !== true) return { outcome: "no-hard-stop" };

  const candidateConnectionIds = candidate.connectionId
    ? [candidate.connectionId]
    : (candidate.allowedConnectionIds ?? []);

+  // No account at all to check. Reporting `state-unknown` here would send the
+  // operator hunting a quota lookup that was never attempted; this is a wiring
+  // problem, not a quota one.
+  if (candidateConnectionIds.length === 0) return { outcome: "no-connection" };
+
  const safe: string[] = [];
+  // An observed exhaustion outranks a missing reading: one is a fact, the other
+  // is the absence of one, and the operator needs the fact. Without this rule the
+  // reason would depend on the order the connections happen to be listed in.
+  let sawExhausted = false;
  for (const connectionId of candidateConnectionIds) {
    if (connectionId === SYNTHETIC_NOAUTH_CONNECTION_ID) continue; // never reachable here, defensive
-    if (isConnectionStateSafe(candidate.provider, connectionId, resolveFreeAccessState, options)) {
-      safe.push(connectionId);
-    }
+    const state = classifyConnectionState(
+      candidate.provider,
+      connectionId,
+      resolveFreeAccessState,
+      options
+    );
+    if (state === "safe") safe.push(connectionId);
+    else if (state === "exhausted") sawExhausted = true;
  }
-  return safe;
+  if (safe.length > 0) return { outcome: "safe", safeConnectionIds: safe };
+  return { outcome: sawExhausted ? "exhausted" : "state-unknown" };
 }

 /**
--- a/open-sse/services/autoCombo/virtualFactory.ts
+++ b/open-sse/services/autoCombo/virtualFactory.ts
@@ -37,7 +37,12 @@ import {
  orderPoolByRung,
  type LadderOptions,
 } from "./subscriptionLadder";
-import { filterStrictZeroCostCandidates, filterTosAvoidCandidates } from "./strictZeroCostFilter";
+import {
+  classifyStrictZeroCostCandidate,
+  filterStrictZeroCostCandidates,
+  filterTosAvoidCandidates,
+  findBudgetEntry,
+} from "./strictZeroCostFilter";
 import { resolveFreeAccessState } from "./freeAccessQuota";
 import { isModelExcludedByConnection } from "@/domain/connectionModelRules";
 import { resolveProviderAlias } from "../model.ts";
@@ -106,6 +111,13 @@ export interface VirtualAutoComboCandidate {
  resolvedSupportsVision?: boolean;
  resolvedReasoning?: boolean;
  resolvedSupportsThinking?: boolean;
+  /**
+   * Why STRICT_ZERO_COST would exclude this candidate, or null when it would
+   * not. Only populated for the read-only inspector build (`skip`), where the
+   * guard is deliberately not applied — dispatch builds leave it undefined and
+   * do no extra work.
+   */
+  freeAccessExclusion?: import("./strictZeroCostFilter").StrictZeroCostExclusionReason | null;
 }

 type VirtualAutoCombo = AutoComboConfig & {
@@ -119,6 +131,9 @@ type VirtualAutoCombo = AutoComboConfig & {
    allowedConnectionIds?: string[];
    weight: number;
    label: string;
+    /** Carried through from the candidate for the read-only inspector; absent
+     * on every dispatch build. */
+    freeAccessExclusion?: import("./strictZeroCostFilter").StrictZeroCostExclusionReason | null;
  }>;
  /** MAX of candidates' context windows — safe to advertise because the
   * auto-combo context pre-filter routes oversized requests to large-window
@@ -749,17 +764,42 @@ export async function prepareVirtualAutoComboInputs(
    // per-candidate: `resolveFreeAccessState` here is a raw pass-through of the real
    // per-(provider,connectionId) resolver; the filter itself decides which connection(s)
    // on each candidate to check and rewrites `allowedConnectionIds` to the SAFE subset.
-    const strictFilteredPool = filterStrictZeroCostCandidates(pool, {
-      enabled: settings.freeAccessPolicy === "strict",
-      resolveFreeAccessState,
+    const strictZeroCostThresholds = {
      // 1 percentage point of headroom, not 0: `freeAccessQuota.ts` reports
      // remaining allowance as a percentage, and a raw ">0" comparison would
      // let a reading of e.g. 0.3% (rounding noise, not real headroom) pass.
      minRemainingAllowance: 1,
      maxStateAgeMs: toNumber(settings.autoRefreshProviderQuotaInterval, 180) * 1000,
+    };
+    const strictZeroCostOn = settings.freeAccessPolicy === "strict";
+    const strictFilteredPool = filterStrictZeroCostCandidates(pool, {
+      // The read-only candidate inspector (#9133) must be able to see what the
+      // guard would exclude, and why — the same opt-out the resilience filter
+      // already honours through `skip`. Dispatch (`skip === false`) is unaffected.
+      enabled: strictZeroCostOn && !skip,
+      resolveFreeAccessState,
+      ...strictZeroCostThresholds,
    });
    if (strictFilteredPool !== pool) pool = strictFilteredPool;

+    // Annotate here rather than in the handler: this is where the thresholds and
+    // `resolveFreeAccessState` already live. Doing it downstream would mean a second
+    // copy of both, with nothing to keep them in agreement.
+    if (strictZeroCostOn && skip) {
+      pool = pool.map((candidate) => {
+        const verdict = classifyStrictZeroCostCandidate(
+          candidate,
+          findBudgetEntry(candidate),
+          resolveFreeAccessState,
+          strictZeroCostThresholds
+        );
+        return {
+          ...candidate,
+          freeAccessExclusion: verdict.outcome === "safe" ? null : verdict.outcome,
+        };
+      });
+    }
+
    // Separate, optional ToS guard — independent of economic safety on purpose.
    const tosFilteredPool = filterTosAvoidCandidates(pool, settings.excludeTosAvoid === true);
    if (tosFilteredPool !== pool) pool = tosFilteredPool;
@@ -1058,6 +1098,9 @@ export async function createVirtualAutoComboFromPrepared(
      : {}),
    weight: snapshotScores.get(candidate.modelStr) ?? 1,
    label: candidate.provider,
+    ...(candidate.freeAccessExclusion === undefined
+      ? {}
+      : { freeAccessExclusion: candidate.freeAccessExclusion }),
  }));
  const autoConfig = {
    candidatePool: providerPool,
--- a/open-sse/services/claudeTlsClient.ts
+++ b/open-sse/services/claudeTlsClient.ts
@@ -2,8 +2,8 @@
 * Browser-TLS-impersonating HTTP client for claude.ai.
 *
 * Thin re-export over the shared `tlsClientBase.ts` factory
- * (`createTlsClientModule`). All provider-agnostic logic (wreq-js transport
- * pooling, direct streaming, proxy resolution, deadlines, SSE detection) lives
+ * (`createTlsClientModule`). All provider-agnostic logic (sidecar lifecycle,
+ * streaming tail-file, proxy resolution, error classes, SSE detection) lives
 * in the base module; this file supplies only Claude-specific config and
 * preserves the original public export surface.
 */
@@ -24,13 +24,13 @@ const HARD_TIMEOUT_GRACE_MS =
 export const tlsClientModule = createTlsClientModule({
  providerName: "Claude",
  tlsProfile: `chrome_${CLAUDE_TLS_BROWSER_MAJOR_VERSION}`,
-  emulationOs: "linux",
  domain: "https://claude.ai",
-  streamEofPolicy: "include",
+  tempDirPrefix: "cgpt-stream-",
+  tailFileVariant: "A",
  responseValidation: "sse",
  exportCloudflareCheck: false,
  exposeStreamingForTesting: true,
-  // Claude allows the native/hard request deadline to bound a slow first SSE byte.
+  // Claude waits indefinitely for the first SSE byte (original 2-arg waitForContent).
  defaultTimeoutMs: DEFAULT_TIMEOUT_MS,
  hardTimeoutGraceMs: HARD_TIMEOUT_GRACE_MS,
  firstByteTimeoutMs: Number.POSITIVE_INFINITY,
--- a/open-sse/services/claudeTurnstileSolver.ts
+++ b/open-sse/services/claudeTurnstileSolver.ts
@@ -7,7 +7,7 @@
 * 3. Waits for Turnstile challenge to appear
 * 4. Waits for challenge to be solved (with retry)
 * 5. Extracts cf_clearance cookie
- * 6. Returns a fresh cookie for the isolated wreq-js request
+ * 6. Returns fresh cookie for tls-client-node
 */

 import type { Browser, Page } from "playwright";
--- a/Show More
+++ b/Show More
				`@@ -0,0 +1 @@`
				- feat(rankings): order Free Provider Rankings by what each provider actually served — `GET /api/free-provider-rankings?sortBy=reliability` and a "Most reliable first" toggle on the page. Providers with too few calls to state a success rate keep their score order below the measured ones; the default order is unchanged ([#12218](https://github.com/diegosouzapw/OmniRoute/pull/12218)).
				`@@ -0,0 +1 @@`
				- feat(auto-combo): Auto-Combo scoring can now weigh how often a provider/model has actually succeeded. The engine already carried that number on every candidate — 24 hours of usage history behind a ten-sample floor, real-time metrics otherwise — and the scoring function never read it, while the weight table described `stability` as if it did. `reliability` (`1 - failureRate`, with the same field precedence and the same rate-bounding the speed ranking already uses, so a corrupt reading means "nothing observed" rather than "fails every call") is now a declared factor shipping at weight `0`, so routing is unchanged until an operator gives it one, and the `stability` description now matches what that factor computes ([#12317](https://github.com/diegosouzapw/OmniRoute/pull/12317))
				`@@ -0,0 +1 @@`
				- feat(routing): With `freeAccessPolicy: "strict"`, the read-only candidate listing (`GET /v1/auto-combo/{channel}/candidates`) no longer hides the candidates the zero-cost guard excludes — the same read-only transparency the resilience filter already honours (#9133). Each candidate now carries `freeAccessExclusion` saying why it would be kept out, and it tells an exhausted allowance apart from a quota reading that never arrived or went stale, which used to look identical from the outside. Routing is unchanged: the listing reports, it never enforces. The separate `excludeTosAvoid` guard still drops its candidates without a reason; that gap is now documented rather than closed ([#12319](https://github.com/diegosouzapw/OmniRoute/pull/12319))
				`@@ -0,0 +1 @@`
				- feat(radar): The Radar catalog table now shows two facts it was already receiving from the feed and dropping on the floor: the per-model rate limits (requests and tokens, per minute and per day) in a new column, and a badge when a provider's terms state it may train on the prompts you send. A limit of zero renders as zero rather than "rate-only" — for a ceiling those are opposite facts — and a model with no training statement gets no badge, because an absent statement is not a guarantee ([#12320](https://github.com/diegosouzapw/OmniRoute/pull/12320))
				`@@ -0,0 +1 @@`
				- fix(sse): passthrough streams now estimate usage on finish when upstream closes without usage even with `stream_options.include_usage` — avoids `0 tokens / 0%` for providers that stay silent (and correctly handles trailing empty-choices usage) ([#12151](https://github.com/diegosouzapw/OmniRoute/pull/12151))
				`@@ -0,0 +1 @@`
				- fix(free-tier): `/api/free-tier/summary` no longer computes its totals from a Radar feed built before the catalog the running release ships. When the cached feed is older — or carries no build date at all — the route answers from the shipped catalog, resolved through the operator's local model state so disabled and tombstoned models stay out of the numbers ([#12215](https://github.com/diegosouzapw/OmniRoute/pull/12215)).
				`@@ -0,0 +1 @@`
				- fix(resilience): a 402 on a single paid model of a passthrough/gateway provider (e.g. `kilo-gateway`, `ollama-cloud`) no longer terminalizes the whole connection with a never-auto-recovered `credits_exhausted` status — only the paid model is locked out, so free models on the same key keep serving. 402 variant of [#3027](https://github.com/diegosouzapw/OmniRoute/issues/3027). Single-credential providers are unaffected — a 402 there is still treated as the key being genuinely out of credit ([#5239](https://github.com/diegosouzapw/OmniRoute/issues/5239) / [#10616](https://github.com/diegosouzapw/OmniRoute/issues/10616)) ([#12242](https://github.com/diegosouzapw/OmniRoute/issues/12242)) — thanks @brick30llc-ctrl
				`@@ -0,0 +1 @@`
				- fix(combo): Expose the two Auto-Combo scoring factors nobody could set — the combo validation schema and the dashboard weight sliders both declared 13 of the scorer's 15 factors, so `connectionDensity` (spreads load across a provider's connections) and `quality` were dropped on save and offered nowhere. The sliders also shipped their own default table that differed from the engine's on every non-zero factor and summed to 1.05, so the percentages shown next to them added up to 105%. Both lists now match `DEFAULT_WEIGHTS`, and a test keeps them there. Note that a combo whose stored `weights` omitted the two keys was effectively running with them at zero and the other thirteen renormalized upward; it now runs with the engine's intended distribution, so its routing does shift ([#12314](https://github.com/diegosouzapw/OmniRoute/pull/12314))
				`@@ -0,0 +1 @@`
				- fix(docs): The free-tier reference no longer says its numbers come "confidence tagged per row" — no catalog entry carries a confidence tag and the API serves none, so every figure on that page is an estimate of the same, unstated quality. The page now states what an entry does vouch for: an independently documented hard stop (set by hand with the source in a comment, never defaulted to `true`) and a prompt-training disclosure, both with live counts the `check:docs-counts` gate keeps honest ([#12318](https://github.com/diegosouzapw/OmniRoute/pull/12318))
				`@@ -0,0 +1 @@`
				- fix(usage): `adobe-firefly` and `firefly` have had a working usage fetcher since Adobe Firefly landed, but neither was ever added to the registration list, so the provider-plugin manifest, `genericQuotaFetcher` and the free-access quota cache all reported them as having no usage support — while `USAGE_SUPPORTED_PROVIDERS` said the opposite. Both are now declared, which also means their credit balance is fetched like any other declared provider's: `registerGenericQuotaFetchers` now registers a generic quota fetcher for them, and `resolveFreeAccessState` no longer returns early. A test holds the registration list to the dispatcher's switch in both directions, which is what the module's own docstring already asked for in prose ([#12321](https://github.com/diegosouzapw/OmniRoute/pull/12321))