coderouter-cli 2.3.0a4__tar.gz → 2.5.0__tar.gz

{coderouter_cli-2.3.0a4 → coderouter_cli-2.5.0}/.gitignore

@@ -32,6 +32,11 @@ env/
 ENV/
 .python-version-local
 
+# ============================================================
+# _OUTPUTS
+# ============================================================
+_OUTPUTS/
+
 # uv
 # Note: uv.lock SHOULD be committed (per plan.md §5.4)
 # Do NOT add uv.lock here.

{coderouter_cli-2.3.0a4 → coderouter_cli-2.5.0}/CHANGELOG.md

@@ -6,6 +6,130 @@ versioning follows [SemVer](https://semver.org/).
 
 ---
 
+## [v2.5.0] — 2026-05-22 (Launcher — llama.cpp / vllm GUI)
+
+Browser-based process manager for local inference backends, integrated
+into the existing CodeRouter web UI at `/launcher`.
+
+### Added
+
+- **`coderouter/ingress/launcher_routes.py`**: New route module providing
+  the Launcher UI and its backing API.
+
+  - `GET /launcher` — Single-page HTML UI (Tailwind CDN + inline JS,
+    same dark-theme aesthetic as `/dashboard`).
+  - `GET /api/launcher/models` — Recursively scans `launcher.model_dirs`
+    and returns discovered model files with name, path, size (GB), and
+    extension.
+  - `GET /api/launcher/option-profiles` — Returns named option presets
+    from `providers.yaml` keyed by backend (`llama.cpp`, `vllm`).
+  - `GET /api/launcher/processes` — Lists all managed processes.
+  - `POST /api/launcher/start` — Starts a backend process. Accepts name,
+    backend, model_path, port, options dict, and extra_args free-text.
+  - `POST /api/launcher/stop/{id}` — SIGTERM → SIGKILL (5 s timeout).
+  - `DELETE /api/launcher/processes/{id}` — Removes a stopped process.
+  - `GET /api/launcher/logs/{id}` — Returns last N lines from the
+    process's 200-line stdout/stderr ring buffer.
+
+- **`LauncherOptionProfile` / `LauncherConfig`** (`config/schemas.py`):
+  New Pydantic models for the `launcher:` block in `providers.yaml`.
+  Adding new CLI flags requires only a YAML edit — no code change.
+
+- **`launcher_profiles.yaml.example`**: Template with 7 llama.cpp
+  presets and 7 vllm presets. For GitHub distribution and community
+  profile contributions.
+
+- **`/dashboard` header**: Added a "Launcher" navigation link.
+
+### Design notes
+
+- **YAML-driven**: option profiles live entirely in `providers.yaml`.
+  No code changes needed to add new backend flags.
+- **Multi-process**: each launched process gets a UUID-based ID and is
+  tracked independently. llama.cpp and vllm can run side by side.
+- **Zero new dependencies**: uses `asyncio.create_subprocess_exec`
+  (stdlib only). The 5-dep invariant is maintained.
+- **In-memory registry**: does not persist across CodeRouter restarts
+  (intentional — avoids zombie GPU allocations on restart).
+
+---
+
+## [v2.4.0] — 2026-05-15 (Goal-session awareness — P1-4/5/6)
+
+Stable release following v2.3.0a4. Promotes the Plugin SDK to stable,
+adds three goal-session features, and ships a rule-suggestion CLI.
+
+### Added
+
+- **`coderouter/guards/_fingerprint.py`** (P1-4): Response fingerprinting
+  helper.  `fingerprint_response(text)` returns a 12-hex SHA-256 digest
+  of the top-N content words (stop-word-filtered, order-independent).
+  Used by the new `goal_progress_stall` drift signal to detect when a
+  model repeats itself without making progress.
+
+- **Signal 6 — `goal_progress_stall`** (`drift_detection.py`, P1-4):
+  Sixth drift signal added to `detect_drift()`.  Fires (mild) when the
+  fraction of fingerprinted responses that repeat an already-seen
+  fingerprint exceeds `repetition_rate_threshold` (default 0.4).
+  Requires `response_fingerprint` to be populated on observations; when
+  absent the signal is silently skipped (backward-compatible).
+
+- **`DriftThresholds.repetition_rate_threshold`** (P1-4): New field on
+  `DriftThresholds`, present on all three presets.  `THRESHOLDS_GOAL`
+  preset added (`min_window_fill=4`, `repetition_rate_threshold=0.2`,
+  tighter across the board) and exposed via `SENSITIVITY_PRESETS["goal"]`.
+
+- **`FallbackChain.goal_mode: bool = False`** (`config/schemas.py`, P1-5):
+  Profile-level flag.  When `True`, the drift detector ignores
+  `drift_detection_sensitivity` and uses `THRESHOLDS_GOAL` instead
+  (stricter thresholds + `min_window_fill=4`).  Designed for `/goal`
+  agent sessions where forward-progress stalls are more actionable.
+
+- **`coderouter/state/suggest_rules.py`** (P1-6): Statistical rule
+  suggestion engine.  `suggest_rules(WindowSummary) → list[RuleSuggestion]`
+  analyses the request journal and emits copy-paste YAML snippets.
+  Five rules: `provider_reorder` (cost rank), `enable_prompt_cache`
+  (high-token / low-hit providers), `enable_drift_detection` (reminder),
+  `low_sensitivity_small_window` (sparse-traffic guard), `goal_profile`
+  (output-divergence → `goal_mode: true`).  Pure statistics — no LLM.
+
+- **`coderouter replay --suggest-rules`** (`cli.py`, P1-6): New flag on
+  the existing `replay` subcommand.  Reads the full request journal,
+  runs `suggest_rules`, and prints a formatted terminal report with
+  confidence badges and YAML snippets.
+
+### Changed
+
+- **`ResponseObservation.response_fingerprint: str | None = None`**
+  (`drift_detection.py`): New optional field (slots-safe, defaults to
+  `None`).  Fully backward-compatible — existing callers that don't
+  populate it get the same five-signal behaviour as before.
+
+- **`FallbackEngine._observe_drift_signal`** (`fallback.py`): Accepts
+  new `response_fingerprint` kwarg.  Non-streaming and streaming success
+  paths now compute and pass a fingerprint for the `goal_progress_stall`
+  signal.  `goal_mode` check applies `THRESHOLDS_GOAL` when the profile
+  flag is set.
+
+### Files touched
+
+```
+A  coderouter/guards/_fingerprint.py
+M  coderouter/guards/__init__.py          — module registry comment
+M  coderouter/guards/drift_detection.py   — Signal 6, THRESHOLDS_GOAL, new fields
+M  coderouter/config/schemas.py           — FallbackChain.goal_mode
+M  coderouter/routing/fallback.py         — fingerprint wiring, goal_mode dispatch
+A  coderouter/state/suggest_rules.py
+M  coderouter/state/__init__.py           — module registry comment
+M  coderouter/cli.py                      — replay --suggest-rules
+A  docs/articles/v1-saga/note-14-v0-4-goal-mode.md
+M  docs/articles/v1-saga/INDEX.md
+M  docs/inside/future.md
+M  CHANGELOG.md, pyproject.toml          — 2.3.0a4 → 2.4.0
+```
+
+---
+
 ## [v2.3.0a4] — 2026-05-08 (Plugin SDK — ruff cleanup)
 
 Patch over `v2.3.0a3`. CI's `ruff check .` job surfaced six lint

{coderouter_cli-2.3.0a4 → coderouter_cli-2.5.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: coderouter-cli
-Version: 2.3.0a4
+Version: 2.5.0
 Summary: Local-first, free-first, fallback-built-in LLM router. Claude Code / OpenAI compatible.
 Project-URL: Homepage, https://github.com/zephel01/CodeRouter
 Project-URL: Repository, https://github.com/zephel01/CodeRouter
@@ -47,7 +47,7 @@ Description-Content-Type: text/markdown
 
 <p align="center">
   <a href="https://github.com/zephel01/CodeRouter/actions/workflows/ci.yml"><img src="https://github.com/zephel01/CodeRouter/actions/workflows/ci.yml/badge.svg?branch=main" alt="CI"></a>
-  <a href=""><img src="https://img.shields.io/badge/version-2.2.0-blue" alt="version"></a>
+  <a href="https://pypi.org/project/coderouter-cli/"><img src="https://img.shields.io/pypi/v/coderouter-cli?include_prereleases&color=blue&label=pypi" alt="pypi"></a>
   <a href=""><img src="https://img.shields.io/badge/python-3.12%2B-blue" alt="python"></a>
   <a href=""><img src="https://img.shields.io/badge/deps-5-brightgreen" alt="deps"></a>
   <a href=""><img src="https://img.shields.io/badge/license-MIT-yellow" alt="license"></a>
@@ -141,7 +141,7 @@ ANTHROPIC_BASE_URL=http://localhost:8088 ANTHROPIC_AUTH_TOKEN=dummy claude
 | ガード | 何から守るか |
 |---|---|
 | **Context Budget** | メッセージが溜まりすぎて context window 溢れ → 自動 trim |
-| **Drift Detection** | モデルの応答品質が徐々に劣化 → 別 provider に切替 or KV cache flush |
+| **Drift Detection** | モデルの応答品質が徐々に劣化 → 別 provider に切替 or KV cache flush (6 シグナル、`goal_mode` で目標達成停滞も検知) |
 | **Self-healing** | backend が落ちた → 自動除外 + restart + 回復 probe で自動復帰 |
 | **Tool Loop Guard** | 同じツールを無限に呼び続ける → 検知して停止 |
 | **Memory Pressure** | GPU メモリ不足を検知 → 軽量モデルに切替 |
@@ -154,9 +154,40 @@ ANTHROPIC_BASE_URL=http://localhost:8088 ANTHROPIC_AUTH_TOKEN=dummy claude
 | **`coderouter doctor`** | プロバイダの問題を 6 プローブで即診断 + 修正パッチ出力 |
 | **`/dashboard`** | ブラウザで今何が起きてるかリアルタイム確認 |
 | **`coderouter audit`** | guard 発火履歴を検索 |
-| **`coderouter replay`** | provider 切替の効果を統計比較 (A/B 分析) |
+| **`coderouter replay`** | provider 切替の効果を統計比較 (A/B 分析) / `--suggest-rules` でルール最適化提案 |
 | **Continuous Probe** | idle 時も定期的に backend を監視 |
 
+### Launcher — llama.cpp / vllm 起動 UI
+
+`http://localhost:8088/launcher` で開けるブラウザ UI。llama.cpp や vllm を GUI で起動・管理できます。
+
+| 機能 | 詳細 |
+|---|---|
+| **モデルスキャン** | `model_dirs` に指定したフォルダを再帰スキャンして `.gguf` / `.safetensors` をリスト化 |
+| **オプションプロファイル** | `providers.yaml` に名前付きプリセットを定義 → ドロップダウンで選択するだけ |
+| **複数プロセス管理** | llama.cpp と vllm を同時に起動し、ポートごとに独立管理 |
+| **ログビューア** | 各プロセスの stdout/stderr をブラウザ内でリアルタイム確認 |
+
+```yaml
+# providers.yaml に追記するだけで有効になる
+launcher:
+  model_dirs:
+    - ~/models
+  option_profiles:
+    llama.cpp:
+      - name: "GPU フル活用"
+        args:
+          "-ngl": 99
+          "--ctx-size": 4096
+    vllm:
+      - name: "標準"
+        args:
+          "--dtype": "auto"
+          "--max-model-len": 4096
+```
+
+詳細 → [Launcher ガイド](./docs/launcher.md)
+
 ---
 
 ## 設定例 (最小)
@@ -193,6 +224,7 @@ providers:
 | すぐ動かす | [Quickstart](./docs/quickstart.md) |
 | 使いこなす | [利用ガイド](./docs/usage-guide.md) |
 | 無料で回す | [無料枠ガイド](./docs/free-tier-guide.md) |
+| llama.cpp / vllm を GUI で起動 | [Launcher ガイド](./docs/launcher.md) |
 | 詰まった | [トラブルシューティング](./docs/troubleshooting.md) |
 | 設計を知りたい | [アーキテクチャ詳細](./docs/architecture.md) |
 | 全リリース履歴 | [CHANGELOG](./CHANGELOG.md) |

{coderouter_cli-2.3.0a4 → coderouter_cli-2.5.0}/README.en.md

@@ -6,7 +6,7 @@
 
 <p align="center">
   <a href="https://github.com/zephel01/CodeRouter/actions/workflows/ci.yml"><img src="https://github.com/zephel01/CodeRouter/actions/workflows/ci.yml/badge.svg?branch=main" alt="CI"></a>
-  <a href=""><img src="https://img.shields.io/badge/version-2.2.0-blue" alt="version"></a>
+  <a href="https://pypi.org/project/coderouter-cli/"><img src="https://img.shields.io/pypi/v/coderouter-cli?include_prereleases&color=blue&label=pypi" alt="pypi"></a>
   <a href=""><img src="https://img.shields.io/badge/python-3.12%2B-blue" alt="python"></a>
   <a href=""><img src="https://img.shields.io/badge/deps-5-brightgreen" alt="deps"></a>
   <a href=""><img src="https://img.shields.io/badge/license-MIT-yellow" alt="license"></a>
@@ -100,7 +100,7 @@ Full decision matrix → [Do I need CodeRouter?](./docs/when-do-i-need-coderoute
 | Guard | What it protects against |
 |---|---|
 | **Context Budget** | Messages piling up → context window overflow. Auto-trim at 90% |
-| **Drift Detection** | Model quality degrading over time → switch provider or flush KV cache |
+| **Drift Detection** | Model quality degrading over time → switch provider or flush KV cache (6 signals incl. `goal_progress_stall`; `goal_mode` for tighter thresholds) |
 | **Self-healing** | Backend crashes → auto-exclude + restart + recovery probe → auto-restore |
 | **Tool Loop Guard** | Agent calling the same tool forever → detect and break |
 | **Memory Pressure** | GPU running out of VRAM → switch to lighter model |
@@ -113,9 +113,40 @@ Full decision matrix → [Do I need CodeRouter?](./docs/when-do-i-need-coderoute
 | **`coderouter doctor`** | 6-probe diagnosis of provider issues + copy-paste YAML patches |
 | **`/dashboard`** | Real-time browser view of what's happening |
 | **`coderouter audit`** | Search guard activation history |
-| **`coderouter replay`** | Compare providers statistically (A/B analysis) |
+| **`coderouter replay`** | Compare providers statistically (A/B analysis) / `--suggest-rules` for automated rule suggestions |
 | **Continuous Probe** | Background health monitoring even during idle |
 
+### Launcher — llama.cpp / vllm GUI
+
+Browser UI at `http://localhost:8088/launcher` for starting and managing local inference backends.
+
+| Feature | Detail |
+|---|---|
+| **Model scanner** | Recursively scans `model_dirs` for `.gguf` / `.safetensors` and lists them |
+| **Option profiles** | Name your flag presets in `providers.yaml` — select from a dropdown, no CLI needed |
+| **Multi-process** | Run llama.cpp and vllm side by side on different ports |
+| **Log viewer** | stdout/stderr of each process shown live in the browser |
+
+```yaml
+# Add to providers.yaml — no code changes needed
+launcher:
+  model_dirs:
+    - ~/models
+  option_profiles:
+    llama.cpp:
+      - name: "Full GPU"
+        args:
+          "-ngl": 99
+          "--ctx-size": 4096
+    vllm:
+      - name: "Standard"
+        args:
+          "--dtype": "auto"
+          "--max-model-len": 4096
+```
+
+Details → [Launcher guide](./docs/launcher.md)
+
 ---
 
 ## Minimal Config
@@ -152,6 +183,7 @@ More detail → [Usage guide](./docs/usage-guide.en.md) · [Architecture](./docs
 | Get running fast | [Quickstart](./docs/quickstart.en.md) |
 | Use it well | [Usage guide](./docs/usage-guide.en.md) |
 | Run for free | [Free-tier guide](./docs/free-tier-guide.en.md) |
+| Launch llama.cpp / vllm via GUI | [Launcher guide](./docs/launcher.md) |
 | Stuck? | [Troubleshooting](./docs/troubleshooting.en.md) |
 | Understand the design | [Architecture](./docs/architecture.md) |
 | Full release history | [CHANGELOG](./CHANGELOG.md) |

{coderouter_cli-2.3.0a4 → coderouter_cli-2.5.0}/README.md

@@ -6,7 +6,7 @@
 
 <p align="center">
   <a href="https://github.com/zephel01/CodeRouter/actions/workflows/ci.yml"><img src="https://github.com/zephel01/CodeRouter/actions/workflows/ci.yml/badge.svg?branch=main" alt="CI"></a>
-  <a href=""><img src="https://img.shields.io/badge/version-2.2.0-blue" alt="version"></a>
+  <a href="https://pypi.org/project/coderouter-cli/"><img src="https://img.shields.io/pypi/v/coderouter-cli?include_prereleases&color=blue&label=pypi" alt="pypi"></a>
   <a href=""><img src="https://img.shields.io/badge/python-3.12%2B-blue" alt="python"></a>
   <a href=""><img src="https://img.shields.io/badge/deps-5-brightgreen" alt="deps"></a>
   <a href=""><img src="https://img.shields.io/badge/license-MIT-yellow" alt="license"></a>
@@ -100,7 +100,7 @@ ANTHROPIC_BASE_URL=http://localhost:8088 ANTHROPIC_AUTH_TOKEN=dummy claude
 | ガード | 何から守るか |
 |---|---|
 | **Context Budget** | メッセージが溜まりすぎて context window 溢れ → 自動 trim |
-| **Drift Detection** | モデルの応答品質が徐々に劣化 → 別 provider に切替 or KV cache flush |
+| **Drift Detection** | モデルの応答品質が徐々に劣化 → 別 provider に切替 or KV cache flush (6 シグナル、`goal_mode` で目標達成停滞も検知) |
 | **Self-healing** | backend が落ちた → 自動除外 + restart + 回復 probe で自動復帰 |
 | **Tool Loop Guard** | 同じツールを無限に呼び続ける → 検知して停止 |
 | **Memory Pressure** | GPU メモリ不足を検知 → 軽量モデルに切替 |
@@ -113,9 +113,40 @@ ANTHROPIC_BASE_URL=http://localhost:8088 ANTHROPIC_AUTH_TOKEN=dummy claude
 | **`coderouter doctor`** | プロバイダの問題を 6 プローブで即診断 + 修正パッチ出力 |
 | **`/dashboard`** | ブラウザで今何が起きてるかリアルタイム確認 |
 | **`coderouter audit`** | guard 発火履歴を検索 |
-| **`coderouter replay`** | provider 切替の効果を統計比較 (A/B 分析) |
+| **`coderouter replay`** | provider 切替の効果を統計比較 (A/B 分析) / `--suggest-rules` でルール最適化提案 |
 | **Continuous Probe** | idle 時も定期的に backend を監視 |
 
+### Launcher — llama.cpp / vllm 起動 UI
+
+`http://localhost:8088/launcher` で開けるブラウザ UI。llama.cpp や vllm を GUI で起動・管理できます。
+
+| 機能 | 詳細 |
+|---|---|
+| **モデルスキャン** | `model_dirs` に指定したフォルダを再帰スキャンして `.gguf` / `.safetensors` をリスト化 |
+| **オプションプロファイル** | `providers.yaml` に名前付きプリセットを定義 → ドロップダウンで選択するだけ |
+| **複数プロセス管理** | llama.cpp と vllm を同時に起動し、ポートごとに独立管理 |
+| **ログビューア** | 各プロセスの stdout/stderr をブラウザ内でリアルタイム確認 |
+
+```yaml
+# providers.yaml に追記するだけで有効になる
+launcher:
+  model_dirs:
+    - ~/models
+  option_profiles:
+    llama.cpp:
+      - name: "GPU フル活用"
+        args:
+          "-ngl": 99
+          "--ctx-size": 4096
+    vllm:
+      - name: "標準"
+        args:
+          "--dtype": "auto"
+          "--max-model-len": 4096
+```
+
+詳細 → [Launcher ガイド](./docs/launcher.md)
+
 ---
 
 ## 設定例 (最小)
@@ -152,6 +183,7 @@ providers:
 | すぐ動かす | [Quickstart](./docs/quickstart.md) |
 | 使いこなす | [利用ガイド](./docs/usage-guide.md) |
 | 無料で回す | [無料枠ガイド](./docs/free-tier-guide.md) |
+| llama.cpp / vllm を GUI で起動 | [Launcher ガイド](./docs/launcher.md) |
 | 詰まった | [トラブルシューティング](./docs/troubleshooting.md) |
 | 設計を知りたい | [アーキテクチャ詳細](./docs/architecture.md) |
 | 全リリース履歴 | [CHANGELOG](./CHANGELOG.md) |

{coderouter_cli-2.3.0a4 → coderouter_cli-2.5.0}/coderouter/cli.py

@@ -293,6 +293,18 @@ def _build_parser() -> argparse.ArgumentParser:
         metavar="N",
         help="Use only the last N entries (applied after --since and --provider filters).",
     )
+    # P1-6: --suggest-rules — statistical analysis → routing rule proposals.
+    replay.add_argument(
+        "--suggest-rules",
+        action="store_true",
+        help=(
+            "P1-6: analyse the request journal and print actionable routing "
+            "rule suggestions as copy-paste YAML snippets. Suggestions cover "
+            "provider reordering by cost, prompt_cache enablement, drift "
+            "detection configuration, and goal profile creation. "
+            "Can be combined with --since / --limit to scope the analysis window."
+        ),
+    )
 
     return parser
 
@@ -684,6 +696,25 @@ def _run_replay(args: argparse.Namespace) -> int:
         print("replay: no matching entries found.")
         return 0
 
+    if getattr(args, "suggest_rules", False):
+        # P1-6: statistical rule suggestion mode.
+        # Always compute a full window summary (ignores --compare / --provider).
+        from coderouter.state.replay import summarize_window as _sw
+        from coderouter.state.suggest_rules import format_suggestions, suggest_rules
+
+        # Re-read without provider filter so we see all providers.
+        all_entries = read_request_log(log_path, since=args.since)
+        if args.limit is not None and args.limit > 0:
+            all_entries = all_entries[-args.limit:]
+        full_summary = _sw(all_entries)
+        suggestions = suggest_rules(full_summary)
+        print(f"Request journal: {len(all_entries)} entries analysed")
+        print(f"  Window: {full_summary.first_ts} → {full_summary.last_ts}")
+        print(f"  Providers: {', '.join(sorted(full_summary.providers))}")
+        print()
+        print(format_suggestions(suggestions))
+        return 0
+
     if args.compare:
         provider_a, provider_b = args.compare
         comparison = compare_providers(entries, provider_a, provider_b)

{coderouter_cli-2.3.0a4 → coderouter_cli-2.5.0}/coderouter/config/schemas.py

@@ -658,6 +658,28 @@ class FallbackChain(BaseModel):
         ),
     )
 
+    # --- P1-5: goal_mode — tighter drift thresholds for /goal sessions -------
+    #
+    # When True, the drift detector automatically switches to the
+    # ``THRESHOLDS_GOAL`` preset regardless of ``drift_detection_sensitivity``,
+    # and lowers ``min_window_fill`` to 4 so stall detection fires faster.
+    #
+    # Intended for profiles routed by the ``/goal`` meta-command where
+    # the agent is expected to make steady forward progress. Repetition and
+    # length collapse are much more meaningful signals in that context than
+    # in a general-purpose chat session.
+    goal_mode: bool = Field(
+        default=False,
+        description=(
+            "P1-5: when True, automatically applies the ``goal`` drift "
+            "threshold preset (stricter thresholds, lower ``min_window_fill`` "
+            "of 4) for this profile. Overrides ``drift_detection_sensitivity`` "
+            "when drift_detection_action is not ``off``. Designed for "
+            "agent/goal sessions where forward-progress stalls are more "
+            "actionable than in ad-hoc chat."
+        ),
+    )
+
     # --- v2.0-H (L6): Mid-stream partial stitching --------------------------
     #   * ``off``      — discard partial content on mid-stream failure (legacy).
     #   * ``surface``  — return partial content as a truncated-but-valid response.
@@ -852,6 +874,130 @@ class AutoRouterConfig(BaseModel):
     )
 
 
+class LauncherBackendConfig(BaseModel):
+    """Per-backend binary path configuration for the Launcher.
+
+    When ``binary`` is unset, the Launcher falls back to the default
+    executable name (``llama-server`` for llama.cpp, ``python`` for vllm)
+    and relies on ``$PATH`` resolution — which works when the tool is
+    globally installed.  Set ``binary`` when:
+
+    - llama.cpp was built from source (e.g. ``~/llama.cpp/build/bin/llama-server``)
+    - vllm lives in a virtualenv (e.g. ``~/.venv/bin/python``)
+    - Multiple builds coexist and you want to pin a specific one
+
+    Tilde (``~``) and environment variables are expanded at launch time.
+
+    Example::
+
+        backends:
+          llama.cpp:
+            binary: ~/llama.cpp/build/bin/llama-server
+          vllm:
+            binary: ~/.venv/bin/python
+    """
+
+    model_config = ConfigDict(extra="forbid")
+
+    binary: str | None = Field(
+        default=None,
+        description=(
+            "Absolute or ``~``-relative path to the backend executable. "
+            "llama.cpp default: ``llama-server`` (PATH). "
+            "vllm default: ``python`` (PATH). "
+            "Expanded at launch time."
+        ),
+    )
+
+
+class LauncherOptionProfile(BaseModel):
+    """One named option preset for a launcher backend (e.g. llama.cpp / vllm).
+
+    ``args`` maps CLI flag strings to their values.  A bool value of
+    ``True`` means "include the flag without a value" (e.g. ``--no-mmap``);
+    ``False`` means "omit the flag entirely".  All other value types are
+    converted to strings and appended as ``--flag value`` pairs.
+
+    Example::
+
+        name: "GPU速度重視"
+        args:
+          "-ngl": 99
+          "--ctx-size": 4096
+          "--no-mmap": false
+    """
+
+    model_config = ConfigDict(extra="forbid")
+
+    name: str = Field(..., description="Display name shown in the Launcher UI dropdown.")
+    args: dict[str, str | int | float | bool] = Field(
+        default_factory=dict,
+        description=(
+            "CLI flag → value mapping. "
+            "bool True = flag only (no value). "
+            "bool False = omit flag. "
+            "All other types are stringified and passed as '--flag value'."
+        ),
+    )
+
+
+class LauncherConfig(BaseModel):
+    """The ``launcher:`` block in providers.yaml.
+
+    Controls the Launcher UI available at ``/launcher``.
+
+    Example::
+
+        launcher:
+          model_dirs:
+            - ~/models
+            - /data/gguf
+          option_profiles:
+            llama.cpp:
+              - name: "GPU速度重視"
+                args:
+                  "-ngl": 99
+                  "--ctx-size": 4096
+            vllm:
+              - name: "標準"
+                args:
+                  "--dtype": "auto"
+                  "--max-model-len": 4096
+    """
+
+    model_config = ConfigDict(extra="forbid")
+
+    model_dirs: list[str] = Field(
+        default_factory=list,
+        description=(
+            "Directories to scan for model files "
+            "(.gguf, .safetensors, .bin, .pt, .ggml). "
+            "Paths are expanded (~ and env vars) at scan time, not at load. "
+            "Non-existent paths are silently skipped."
+        ),
+    )
+    backends: dict[str, LauncherBackendConfig] = Field(
+        default_factory=dict,
+        description=(
+            "Per-backend binary path overrides. "
+            "Keys are backend names ('llama.cpp', 'vllm'). "
+            "When a key is absent, the default executable is used "
+            "('llama-server' / 'python') and resolved via PATH. "
+            "Useful when running a from-source build or a venv-specific binary."
+        ),
+    )
+    option_profiles: dict[str, list[LauncherOptionProfile]] = Field(
+        default_factory=dict,
+        description=(
+            "Named option presets per backend. "
+            "Keys should be backend names: 'llama.cpp', 'vllm'. "
+            "Each key maps to an ordered list of named presets. "
+            "A free-form 'extra args' field is always available in the UI "
+            "for one-off overrides without touching this config."
+        ),
+    )
+
+
 class PluginsConfig(BaseModel):
     """The ``plugins:`` block in providers.yaml (v2.3.0).
 
@@ -1060,6 +1206,17 @@ class CodeRouterConfig(BaseModel):
             "plugins (zero-cost, backward-compatible default)."
         ),
     )
+    launcher: LauncherConfig | None = Field(
+        default=None,
+        description=(
+            "Launcher configuration for the /launcher UI. "
+            "Defines model_dirs to scan and option_profiles per backend "
+            "('llama.cpp', 'vllm'). "
+            "Unset (None) = Launcher UI shows empty model list and no profiles. "
+            "The Launcher UI itself is always available at /launcher "
+            "regardless of this setting."
+        ),
+    )
 
     @model_validator(mode="after")
     def _check_default_profile_exists(self) -> CodeRouterConfig:

{coderouter_cli-2.3.0a4 → coderouter_cli-2.5.0}/coderouter/guards/__init__.py

@@ -12,6 +12,8 @@ to hit:
   * :mod:`coderouter.guards.self_healing`    — v2.0-J auto-exclude +
                                                  restart + recovery probe
   * :mod:`coderouter.guards.continuous_probe` — v2.0-I background probing
+  * :mod:`coderouter.guards._fingerprint`     — P1-4 response fingerprinting
+                                                 for goal_progress_stall signal
 
 Each guard is a pure-functional / single-class module that the engine
 consults at the appropriate dispatch point. Guards never block the