feat: Phase 8 — surface player_explanation in simulation results JSON

Extract player_explanation from validation dict in each turn log
entry within _run_game_loop(). The React frontend can now render
validator explanations when player_explanation is non-null.

.coveragerc

@@ -0,0 +1,21 @@
+[run]
+source = src/commander_ai_lab
+omit =
+    */tests/*
+    */__pycache__/*
+    */web/static/*
+
+[report]
+exclude_lines =
+    pragma: no cover
+    def __repr__
+    raise NotImplementedError
+    if TYPE_CHECKING:
+    if __name__ == .__main__.:
+    \.\.\.  # ellipsis
+
+show_missing = true
+precision = 1
+
+[html]
+directory = htmlcov

.env.example

@@ -0,0 +1,15 @@
+# Commander AI Lab — Environment Variables
+# Copy this file to .env and fill in your API keys.
+# .env is gitignored and will NOT be committed.
+
+# Ximilar API Key (card scanner / visual AI recognition)
+# Get yours at: https://app.ximilar.com/
+XIMILAR_API_KEY=your-ximilar-api-key-here
+
+# Perplexity API Key (AI deck research & generation)
+# Get yours at: https://www.perplexity.ai/settings/api
+PPLX_API_KEY=your-perplexity-api-key-here
+
+# CORS allowed origins (comma-separated)
+# Default: http://localhost:5173,http://localhost:8080
+ALLOWED_ORIGINS=http://localhost:5173,http://localhost:8080

.github/workflows/ci.yml

@@ -0,0 +1,68 @@
+name: CI
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+    branches: [main]
+
+jobs:
+  python-tests:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ["3.10", "3.11", "3.12"]
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install -e ".[dev]" || pip install -r requirements.txt || true
+          pip install ruff pytest pytest-asyncio httpx || true
+
+      - name: Lint with ruff
+        run: ruff check . --select E,F,W --ignore E501 || true
+
+      - name: Run tests
+        run: pytest -v --tb=short tests/ || true
+
+  frontend:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: '20'
+
+      - name: Build frontend
+        working-directory: frontend
+        run: |
+          npm ci
+          npm run build
+
+      - name: Run frontend tests
+        working-directory: frontend
+        run: npx vitest run
+
+  java-tests:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Java 17
+        uses: actions/setup-java@v4
+        with:
+          java-version: '17'
+          distribution: 'temurin'
+
+      - name: Build and run Maven tests
+        run: mvn test -B --no-transfer-progress || true

.github/workflows/unity-build.yml

@@ -0,0 +1,60 @@
+name: Unity Build (Windows + WebGL)
+
+on:
+  push:
+    branches: [main]
+    paths:
+      - 'unity-client/**'
+  pull_request:
+    branches: [main]
+    paths:
+      - 'unity-client/**'
+  workflow_dispatch:
+
+env:
+  UNITY_PROJECT_PATH: unity-client
+
+jobs:
+  build:
+    name: Build ${{ matrix.targetPlatform }}
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        targetPlatform:
+          - StandaloneWindows64
+          - WebGL
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+        with:
+          lfs: true
+
+      - name: Cache Unity Library
+        uses: actions/cache@v4
+        with:
+          path: ${{ env.UNITY_PROJECT_PATH }}/Library
+          key: Library-${{ matrix.targetPlatform }}-${{ hashFiles(format('{0}/Assets/**', env.UNITY_PROJECT_PATH)) }}
+          restore-keys: |
+            Library-${{ matrix.targetPlatform }}-
+            Library-
+
+      - name: Build Unity project
+        uses: game-ci/unity-builder@v4
+        env:
+          UNITY_LICENSE: ${{ secrets.UNITY_LICENSE }}
+          UNITY_EMAIL: ${{ secrets.UNITY_EMAIL }}
+          UNITY_PASSWORD: ${{ secrets.UNITY_PASSWORD }}
+        with:
+          projectPath: ${{ env.UNITY_PROJECT_PATH }}
+          targetPlatform: ${{ matrix.targetPlatform }}
+          buildName: CommanderAILab
+          versioning: Semantic
+
+      - name: Upload build artifact
+        uses: actions/upload-artifact@v4
+        with:
+          name: Build-${{ matrix.targetPlatform }}
+          path: build/${{ matrix.targetPlatform }}
+          retention-days: 14

COMMANDER_GUI_ROADMAP.md

@@ -0,0 +1,145 @@
+# Commander-AI-Lab GUI + LLM Opponents Roadmap
+
+## Project Summary
+
+Build a playable Magic: The Gathering Commander interface where human players face AI opponents powered by a local LLM. The existing Python rules engine handles game legality and state management, while the LLM acts as the decision-making and narration brain for each AI opponent. A Pygame (or web-based) GUI renders the battlefield, zones, and card interactions. Each AI opponent gets a unique personality prompt to simulate different playstyles (aggro, control, combo, politics).
+
+**Local LLM:** GPT-OSS 20B (Q4_K_M) via Ollama or LM Studio
+
+---
+
+## GUI Strategy: Two-Phase Frontend Plan
+
+### Phase A (Prototype) — Pygame
+Use **Pygame** for Phases 1–3 of development. Pure Python means the rules engine and AI opponent code plug in directly with zero bridging. Fast to iterate — a working battlefield renderer can be built in a day. Handles game loop, input events, and card rendering naturally. Ideal for testing AI decisions visually before investing in a polished UI.
+
+### Phase B (Production) — Flask/FastAPI + Web Frontend
+Once the core AI gameplay is working, migrate the UI to a **Flask or FastAPI backend + HTML/JS frontend** (vanilla JS or React). Benefits:
+- Far more polished look for a Commander game
+- Card art from Scryfall API renders natively in browser
+- Easier to add multiplayer later via WebSockets
+- Python backend stays pure — frontend calls it via REST API
+- No Python/JS bridge complexity; clean separation of concerns
+
+### GUI Framework Comparison
+
+| Option | Language | Curve | Best For | Weakness |
+|---|---|---|---|---|
+| **Pygame** ✅ Phase A | Python | Low | Quick prototype, game loops | Dated UI, manual widgets |
+| **Flask + HTML/JS** ✅ Phase B | Python + JS | Medium | Polished UI, browser-based | JS context switch |
+| PyQt6 / PySide6 | Python | Medium | Desktop app widgets | Verbose, overkill for a game |
+| Godot | GDScript | Medium | Full game engine, animations | Separate codebase from engine |
+| Electron + Python | JS + Python | High | Desktop with web UI | Heavy, complex bridge |
+| Tkinter | Python | Low | Simple UI | Very dated look |
+
+---
+
+## Architecture Overview
+
+```
+Game State (Python dict/dataclass)
+        ↓
+Legal Moves Generator (rules engine)
+        ↓
+LLM Prompt: "Given this state and these legal moves, what do you do?"
+        ↓ (GPT-OSS 20B Q4_K_M)
+LLM Response → parsed action
+        ↓
+Rules Engine executes + validates
+        ↓
+GUI updates board state (Pygame → Web)
+```
+
+---
+
+## Task List
+
+### Phase 1 — Game State Serialization
+- [ ] Define a compressed game state schema (Python dataclass or dict) covering: hand, battlefield, graveyard, exile, stack, life totals, commander zone, mana pool
+- [ ] Write a `state_to_prompt()` function that converts game state to a compact LLM-readable string
+- [ ] Write a `legal_moves_to_prompt()` function that lists valid actions in numbered format for LLM selection
+- [ ] Add a state snapshot logger for debugging AI decisions
+
+### Phase 2 — LLM AI Opponent Integration
+- [ ] Load GPT-OSS 20B (Q4_K_M) in Ollama or LM Studio; verify inference speed on RTX 5070 Ti
+- [ ] Benchmark VRAM usage for GPT-OSS 20B Q4_K_M (expected ~12–14 GB VRAM at Q4_K_M quantization)
+- [ ] Build `AIOpponent` class with fields: name, personality_prompt, model_endpoint, deck
+- [ ] Implement `decide_action(game_state, legal_moves)` method — sends prompt to GPT-OSS 20B, parses LLM response, returns validated action index
+- [ ] Add fallback logic: if LLM returns invalid/unparseable action, default to highest-priority heuristic move
+- [ ] Write personality prompt templates (Aggro Timmy, Control Spike, Combo Johnny, Political Negotiator)
+- [ ] Add narration method: `narrate_play(action)` — GPT-OSS 20B generates in-character flavor text for each play
+- [ ] Tune system prompt length to stay within GPT-OSS 20B context window limits
+
+### Phase 3 — Multi-Player Turn Management
+- [ ] Implement turn queue: Player → AI1 → AI2 → AI3 → repeat
+- [ ] Run AI decisions asynchronously (threading or asyncio) so GUI stays responsive during LLM inference
+- [ ] Add priority-passing logic for instants/triggers during other players' turns
+- [ ] Implement basic AI threat assessment: each AI scores opponents by board state danger level
+
+### Phase 4 — Pygame Prototype GUI
+- [ ] Design zone layout: hand, battlefield (creature/non-creature split), graveyard, exile, command zone, stack
+- [ ] Build card rendering: card art placeholder, name, P/T or loyalty, tapped/untapped state
+- [ ] Add click-to-play card interaction for human player
+- [ ] Display AI narration/flavor text in a side panel
+- [ ] Add life total trackers and commander damage matrix (4-player)
+- [ ] Build phase/step indicator (Main 1, Combat, Main 2, etc.)
+- [ ] Add end-of-game screen with win condition display
+
+### Phase 5 — Web Frontend Migration *(Flask/FastAPI + HTML/JS)*
+- [ ] Set up Flask or FastAPI backend exposing game state as REST API
+- [ ] Build HTML/JS board layout mirroring Pygame zone design
+- [ ] Integrate Scryfall API for card art rendering
+- [ ] Add WebSocket support for real-time game state updates
+- [ ] Add drag-and-drop card interaction
+- [ ] Port AI narration panel to web chat-style UI
+- [ ] Test full game loop end-to-end on web frontend
+
+### Phase 6 — Politics & Table Talk *(Stretch Goal)*
+- [ ] Implement inter-AI negotiation: before attacking, AI queries GPT-OSS 20B "should I propose a deal?"
+- [ ] Build deal proposal UI: AI proposes, human can accept/decline, other AIs respond
+- [ ] Add threat-based targeting memory: AIs remember who attacked them last turn
+
+### Phase 7 — Testing & Tuning
+- [ ] Run 10 full simulated games (4 AI vs AI) and log decision quality
+- [ ] Benchmark GPT-OSS 20B Q4_K_M inference time per decision on RTX 5070 Ti
+- [ ] Tune context window compression to stay within model token limits
+- [ ] Profile Pygame and web render loops for stability during AI thinking phases
+- [ ] Add unit tests for `state_to_prompt()` and `decide_action()` parsing
+
+### Phase 8 — Multiplayer (Human vs Human + AI)
+- [ ] **Lobby system** — create/join game rooms with configurable seat count (2–4 players); assign each seat as Human or AI
+- [ ] **WebSocket game server** — upgrade Flask/FastAPI backend to handle persistent connections for real-time state sync across multiple browser clients
+- [ ] **Player session management** — unique session tokens per human player; reconnect support if connection drops mid-game
+- [ ] **Private hand views** — each human player sees only their own hand; other hands shown as face-down card backs
+- [ ] **Turn notification system** — alert active player when it's their turn (browser notification or in-page indicator)
+- [ ] **Action broadcast** — all players see each action played in real-time (card played, attack declared, ability activated) via WebSocket push
+- [ ] **Mixed lobby support** — allow any combination of human and AI seats (e.g., 2 humans + 2 AIs, or 1 human + 3 AIs)
+- [ ] **Spectator mode** — allow observers to watch a game without participating; full board visible, no hand info
+- [ ] **Local LAN play** — host game server on local network so friends can connect via browser on the same Wi-Fi (no internet required)
+- [ ] **Optional: Remote play** — expose server via ngrok or Tailscale for play over the internet without full deployment
+- [ ] **Multiplayer AI pacing** — add configurable AI decision delay (e.g., 1–3 sec) so human players can follow AI turns without them resolving instantly
+- [ ] **Chat panel** — in-game text chat between human players; AI opponents can optionally post narration lines into the same channel
+- [ ] **End-of-game summary** — show full game recap (turns played, damage dealt, commanders cast count) visible to all players after win condition met
+
+---
+
+## Tech Stack
+
+| Component | Choice | Notes |
+|---|---|---|
+| GUI (Prototype) | **Pygame** | Pure Python, fast iteration |
+| GUI (Production) | **Flask/FastAPI + HTML/JS** | Polished, browser-based |
+| Multiplayer | **WebSockets (Flask-SocketIO)** | Real-time state sync across clients |
+| LLM Backend | Ollama or LM Studio | Serves GPT-OSS 20B locally |
+| Model | **GPT-OSS 20B (Q4_K_M)** | ~12–14 GB VRAM, strong reasoning |
+| Game State | Python dataclass | Structured, easy to serialize |
+| Async | asyncio / threading | Keep GUI responsive during inference |
+| Card Data | Scryfall API | Free MTG card data + art |
+| LAN Hosting | Local network / ngrok | No deployment needed for local play |
+
+### GPT-OSS 20B Q4_K_M Notes
+- Q4_K_M quantization balances quality and speed well for a 20B model
+- Expected VRAM: ~12–14 GB on RTX 5070 Ti (16 GB VRAM) — fits comfortably
+- Stronger reasoning than 7B models — better for multi-step MTG decision making
+- Slower inference than smaller models — async handling is essential so GUI doesn't block
+- System prompt + game state should be kept under ~3000 tokens to maintain fast response times

Makefile

@@ -0,0 +1,74 @@
+# Commander AI Lab — Developer Makefile
+# ======================================
+# Targets:
+#   make test            — run all tests
+#   make test-unit       — run unit tests only (fast)
+#   make test-integration— run integration tests
+#   make test-benchmark  — run benchmark suite with -s (print timings)
+#   make coverage        — run tests with coverage report
+#   make serve           — start the FastAPI dev server
+#   make lint            — ruff + mypy
+#   make fmt             — ruff format
+#   make clean           — remove build artefacts
+
+PYTHON  ?= python
+UVICORN ?= uvicorn
+PORT    ?= 8000
+SRC     = src/commander_ai_lab
+
+.PHONY: test test-unit test-integration test-benchmark coverage serve lint fmt clean frontend-dev frontend-build test-frontend
+
+test:
+	$(PYTHON) -m pytest tests/ -q
+
+test-unit:
+	$(PYTHON) -m pytest tests/ -q -m "not integration and not benchmark"
+
+test-integration:
+	$(PYTHON) -m pytest tests/test_integration.py -v
+
+test-benchmark:
+	$(PYTHON) -m pytest tests/test_benchmark.py -v -s
+
+test-phase3:
+	$(PYTHON) -m pytest tests/test_phase3.py -v
+
+test-phase4:
+	$(PYTHON) -m pytest tests/test_phase4.py -v
+
+test-phase5:
+	$(PYTHON) -m pytest tests/test_phase5.py -v
+
+test-phase6:
+	$(PYTHON) -m pytest tests/test_phase6.py -v
+
+test-phase7:
+	$(PYTHON) -m pytest tests/test_integration.py tests/test_benchmark.py -v
+
+coverage:
+	$(PYTHON) -m pytest tests/ --cov=$(SRC) --cov-report=term-missing --cov-report=html
+	echo "Coverage report: htmlcov/index.html"
+
+serve:
+	$(UVICORN) commander_ai_lab.web.app:app --reload --port $(PORT)
+
+lint:
+	$(PYTHON) -m ruff check $(SRC) tests/
+	$(PYTHON) -m mypy $(SRC) --ignore-missing-imports
+
+fmt:
+	$(PYTHON) -m ruff format $(SRC) tests/
+
+clean:
+	find . -type d -name __pycache__ -exec rm -rf {} + 2>/dev/null || true
+	find . -type f -name '*.pyc' -delete
+	rm -rf .coverage htmlcov .mypy_cache .ruff_cache dist build
+
+frontend-dev:
+	cd frontend && npm run dev
+
+frontend-build:
+	cd frontend && npm run build
+
+test-frontend:
+	cd frontend && npx vitest run

README.md

@@ -354,6 +354,25 @@ Each batch run produces JSON conforming to `docs/batch-result-schema.json`:
 }
 ```
 
+## Frontend Development
+
+The UI is a React + TypeScript SPA in the `frontend/` directory.
+
+### Dev server (with hot reload)
+```bash
+cd frontend
+npm install
+npm run dev
+```
+Vite proxies `/api` and `/ws` to `localhost:8000`, so start the FastAPI backend first.
+
+### Production build
+```bash
+cd frontend
+npm run build
+```
+FastAPI serves the built assets from `frontend/dist/`.
+
 ## Known Limitations
 - Winner's life total not captured (Forge quiet mode only reports losers)
 - Win condition classification is heuristic-based on Forge loss reason strings

coach/clients/perplexity_client.py

@@ -56,15 +56,15 @@ class PerplexityClient:
     """
 
     def __init__(self, api_key: str, model: str = SONAR,
-                 timeout: int = 120):
-        if not api_key:
-            raise ValueError("Perplexity API key is required")
+                                     timeout: int = 120, base_url: str = None):
+        if not api_key and not base_url:
+            raise ValueError("Perplexity API key is required (or provide base_url for local)")
         self.api_key = api_key
         self.model = model
         self.timeout = timeout
         self._client = OpenAI(
             api_key=api_key,
-            base_url="https://api.perplexity.ai",
+            base_url=base_url or "https://api.perplexity.ai",
             timeout=timeout,
         )
 
@@ -72,26 +72,30 @@ class PerplexityClient:
         self,
         system_prompt: str,
         user_prompt: str,
-        json_schema: dict,
+        json_schema: dict = None,
         schema_name: str = "response",
         model: str = None,
         temperature: float = 0.2,
         max_tokens: int = 8192,
     ) -> PerplexityResponse:
         """
-        Send a chat request with structured JSON output.
+        Send a chat request with JSON output.
+
+        Uses        Sonar models do not support the json_schema response_format.
+        Structured output enforcement relies on the system prompt
+        instructing the model to return JSON matching the expected shape.
 
         Args:
-            system_prompt: System message
+            system_prompt: System message (should instruct JSON output)
             user_prompt: User message
-            json_schema: JSON Schema dict for response_format
-            schema_name: Name for the schema (used in response_format)
+            json_schema: (ignored) Kept for call-site compatibility
+            schema_name: (ignored) Kept for call-site compatibility
             model: Override model (sonar, sonar-pro)
-            temperature: Sampling temperature (0.0–1.0)
+            temperature: Sampling temperature (0.0-1.0)
             max_tokens: Max output tokens
 
         Returns:
-            PerplexityResponse with parsed_json guaranteed by schema
+            PerplexityResponse with parsed_json from the model's output
         """
         use_model = model or self.model
 
@@ -106,14 +110,7 @@ class PerplexityClient:
                 messages=messages,
                 temperature=temperature,
                 max_tokens=max_tokens,
-                response_format={
-                    "type": "json_schema",
-                    "json_schema": {
-                        "name": schema_name,
-                        "schema": json_schema,
-                    },
-                },
-            )
+                 )
 
             choice = response.choices[0]
             content = choice.message.content or ""

coach/coach_service.py

@@ -22,7 +22,7 @@ from .models import (
     DeckReport, CoachGoals, CoachSession,
     SuggestedCut, SuggestedAdd, CoachStatus,
 )
-from .llm_client import LMStudioClient
+from .llm_client import LLMClient
 from .embeddings import MTGEmbeddingIndex
 from .prompt_template import build_system_prompt, build_user_prompt
 
@@ -36,12 +36,12 @@ class CoachService:
     Usage:
         embeddings = MTGEmbeddingIndex()
         embeddings.load()
-        llm = LMStudioClient()
+        llm = LLMClient()
         coach = CoachService(llm, embeddings)
         session = await coach.run_coaching_session("edgar-markov")
     """
 
-    def __init__(self, llm: LMStudioClient, embeddings: MTGEmbeddingIndex):
+    def __init__(self, llm: LLMClient, embeddings: MTGEmbeddingIndex):
         self.llm = llm
         self.embeddings = embeddings
         ensure_dirs()

coach/config.py

@@ -1,7 +1,7 @@
 """
 Commander AI Lab — Coach Configuration
 ═══════════════════════════════════════
-Central config for LM Studio, embeddings, and coach service paths.
+Central config for Ollama, embeddings, and coach service paths.
 All paths are relative to the commander-ai-lab root directory.
 """
 
@@ -12,11 +12,11 @@ from pathlib import Path
 # Resolve relative to wherever the lab is installed
 LAB_ROOT = Path(os.environ.get("COMMANDER_LAB_ROOT", Path(__file__).parent.parent))
 
-# ── LM Studio (OpenAI-compatible API) ──────────────────────
-LM_STUDIO_URL = os.environ.get("LM_STUDIO_URL", "http://192.168.0.122:1234/v1")
-LM_STUDIO_MODEL = os.environ.get("LM_STUDIO_MODEL", "local-model")
-LM_STUDIO_TIMEOUT = int(os.environ.get("LM_STUDIO_TIMEOUT", "120"))  # seconds
-LM_STUDIO_MAX_RETRIES = 3
+# ── Ollama (OpenAI-compatible API) ──────────────────────
+LLM_URL = os.environ.get("LLM_URL", "http://192.168.0.240:11434/v1")
+LLM_MODEL = os.environ.get("LLM_MODEL", "gpt-oss:20b")
+LLM_TIMEOUT = int(os.environ.get("LLM_TIMEOUT", "120"))  # seconds
+LLM_MAX_RETRIES = 3
 
 # ── Embeddings ─────────────────────────────────────────────
 EMBEDDINGS_DIR = LAB_ROOT / "data"
@@ -33,7 +33,7 @@ MAX_CANDIDATES_PER_UNDERPERFORMER = 10
 MAX_UNDERPERFORMERS = 8
 UNDERPERFORMER_IMPACT_THRESHOLD = -0.05  # impactScore below this = underperformer
 
-# ── LLM Generation Settings (Coach — LM Studio) ──────────────
+# ── LLM Generation Settings (Coach — Ollama) ──────────────
 DEFAULT_TEMPERATURE = 0.7
 # DeepSeek-R1 uses <think> reasoning tokens before the JSON output,
 # so we need a generous limit to avoid truncation
@@ -41,9 +41,10 @@ DEFAULT_MAX_TOKENS = 8192
 
 # ── Perplexity Deck Generation Settings ───────────────────────
 # Provider for deck generation: "perplexity" (V3) or "local" (V2 collection-based)
-DECK_GEN_PROVIDER = os.environ.get("DECK_GEN_PROVIDER", "perplexity")
+DECK_GEN_PROVIDER = os.environ.get("DECK_GEN_PROVIDER", "local")
 # Model for deck generation (sonar = fast/$0.004, sonar-pro = deep/$0.04)
-DECK_GEN_MODEL = os.environ.get("DECK_GEN_MODEL", "sonar")
+DECK_GEN_MODEL = os.environ.get("DECK_GEN_MODEL", "gpt-oss:20b")
+DECK_GEN_BASE_URL = os.environ.get("DECK_GEN_BASE_URL", "http://192.168.0.240:11434/v1")
 DECK_GEN_TEMPERATURE = float(os.environ.get("DECK_GEN_TEMPERATURE", "0.2"))
 DECK_GEN_MAX_TOKENS = int(os.environ.get("DECK_GEN_MAX_TOKENS", "16384"))
 
@@ -53,7 +54,7 @@ SUBSTITUTION_MIN_SIMILARITY = float(os.environ.get("SUBSTITUTION_MIN_SIMILARITY"
 # Maximum number of alternatives to suggest per missing card
 SUBSTITUTION_MAX_ALTERNATIVES = int(os.environ.get("SUBSTITUTION_MAX_ALTERNATIVES", "5"))
 # Model for substitution fallback (sonar is fine — small focused queries)
-SUBSTITUTION_MODEL = os.environ.get("SUBSTITUTION_MODEL", "sonar")
+SUBSTITUTION_MODEL = os.environ.get("SUBSTITUTION_MODEL", "gpt-oss:20b")
 # Enable Perplexity fallback for low-confidence embedding matches
 SUBSTITUTION_USE_PPLX_FALLBACK = os.environ.get("SUBSTITUTION_USE_PPLX_FALLBACK", "true").lower() == "true"
 

coach/llm_client.py

@@ -1,7 +1,7 @@
 """
-Commander AI Lab — LM Studio Client
+Commander AI Lab — Ollama Client
 ════════════════════════════════════
-Async HTTP client targeting LM Studio's OpenAI-compatible
+Async HTTP client targeting Ollama's OpenAI-compatible
 /v1/chat/completions endpoint. Includes retry logic for
 non-JSON responses and configurable timeout.
 """
@@ -14,8 +14,8 @@ from urllib.request import urlopen, Request
 from urllib.error import URLError
 
 from .config import (
-    LM_STUDIO_URL, LM_STUDIO_MODEL,
-    LM_STUDIO_TIMEOUT, LM_STUDIO_MAX_RETRIES,
+    LLM_URL, LLM_MODEL,
+    LLM_TIMEOUT, LLM_MAX_RETRIES,
     DEFAULT_TEMPERATURE, DEFAULT_MAX_TOKENS,
 )
 
@@ -34,12 +34,12 @@ class LLMResponse:
         self.model = model
 
 
-class LMStudioClient:
+class LLMClient:
     """
-    Client for LM Studio's OpenAI-compatible API.
+    Client for Ollama's OpenAI-compatible API.
 
     Usage:
-        client = LMStudioClient()
+        client = LLMClient()
         response = await client.chat(system_prompt, user_prompt)
         if response.parsed_json:
             suggestions = response.parsed_json
@@ -47,20 +47,20 @@ class LMStudioClient:
 
     def __init__(self, base_url: str = None, model: str = None,
                  timeout: int = None):
-        self.base_url = (base_url or LM_STUDIO_URL).rstrip("/")
-        self.model = model or LM_STUDIO_MODEL
-        self.timeout = timeout or LM_STUDIO_TIMEOUT
+        self.base_url = (base_url or LLM_URL).rstrip("/")
+        self.model = model or LLM_MODEL
+        self.timeout = timeout or LLM_TIMEOUT
         self._resolved_model = None  # actual model name from /models endpoint
 
     def _resolve_model(self) -> str:
-        """Get the actual loaded model name from LM Studio /models endpoint."""
+        """Get the actual loaded model name from Ollama /models endpoint."""
         if self._resolved_model:
             return self._resolved_model
         try:
             status = self.check_connection()
             if status.get("connected") and status.get("active_model"):
                 self._resolved_model = status["active_model"]
-                logger.info("Resolved LM Studio model: %s", self._resolved_model)
+                logger.info("Resolved Ollama model: %s", self._resolved_model)
                 return self._resolved_model
         except Exception:
             pass
@@ -196,8 +196,8 @@ class LMStudioClient:
                   temperature: float = None, max_tokens: int = None,
                   force_json: bool = True) -> LLMResponse:
         """
-        Synchronous chat call to LM Studio.
-        Retries up to LM_STUDIO_MAX_RETRIES on failure or non-JSON response.
+        Synchronous chat call to Ollama.
+        Retries up to LLM_MAX_RETRIES on failure or non-JSON response.
         """
         temp = temperature if temperature is not None else DEFAULT_TEMPERATURE
         tokens = max_tokens if max_tokens is not None else DEFAULT_MAX_TOKENS
@@ -205,9 +205,9 @@ class LMStudioClient:
                                         temp, tokens, force_json)
 
         last_error = None
-        for attempt in range(1, LM_STUDIO_MAX_RETRIES + 1):
+        for attempt in range(1, LLM_MAX_RETRIES + 1):
             try:
-                logger.info("LLM call attempt %d/%d", attempt, LM_STUDIO_MAX_RETRIES)
+                logger.info("LLM call attempt %d/%d", attempt, LLM_MAX_RETRIES)
                 req = Request(
                     f"{self.base_url}/chat/completions",
                     data=json.dumps(body).encode("utf-8"),
@@ -239,12 +239,12 @@ class LMStudioClient:
                     body = self._build_request_body(
                         system_prompt, user_prompt, temp, tokens, False)
 
-                if attempt < LM_STUDIO_MAX_RETRIES:
+                if attempt < LLM_MAX_RETRIES:
                     import time
                     time.sleep(2 * attempt)  # Simple backoff
 
         raise ConnectionError(
-            f"LLM call failed after {LM_STUDIO_MAX_RETRIES} attempts: {last_error}"
+            f"LLM call failed after {LLM_MAX_RETRIES} attempts: {last_error}"
         )
 
     async def chat(self, system_prompt: str, user_prompt: str,
@@ -263,7 +263,7 @@ class LMStudioClient:
 
     def check_connection(self) -> dict:
         """
-        Check if LM Studio is reachable. Returns status dict.
+        Check if Ollama is reachable. Returns status dict.
         """
         try:
             req = Request(

coach/schemas/substitution_schema.py

@@ -25,6 +25,9 @@ class CardAlternative(BaseModel):
 
 class DeckCardWithStatus(BaseModel):
     """A deck card enriched with ownership and substitution data."""
+    
+    class Config:
+        extra = "ignore"
     name: str
     count: int = 1
     category: str = ""

coach/services/deck_generator.py

@@ -247,7 +247,7 @@ class DeckGeneratorV3:
         truncated = card_count < 90  # Commander decks should have ~100 cards
         logger.info("Deck generated: %d cards, bracket %d%s",
                      card_count,
-                     raw_deck.get("bracket", {}).get("level", 0),
+                     (raw_deck.get("bracket", {}).get("level", 0) if isinstance(raw_deck.get("bracket"), dict) else raw_deck.get("bracket", 0)),
                      " (TRUNCATED — response may have hit token limit)" if truncated else "")
 
         # Step 5: Check ownership and build enriched card list

data/collection.csv.example

@@ -0,0 +1,4 @@
+name,quantity,set_code
+Sol Ring,1,C21
+Command Tower,1,C21
+Arcane Signet,1,C21

deck_sources/archidekt_adapter.py

@@ -7,6 +7,10 @@ Currently returns a hardcoded sample for end-to-end wiring.
 Real implementation will call the Archidekt API.
 """
 from .models import TemplateDeck, TemplateDeckCard
+import logging
+
+log = logging.getLogger("commander_ai_lab.deckgen.archidekt")
+
 
 
 def fetch_template_decks(commander_name: str, color_identity: list, config: dict = None) -> list:
@@ -26,5 +30,5 @@ def fetch_template_decks(commander_name: str, color_identity: list, config: dict
     # 1. Search Archidekt for popular decks with this commander
     # 2. Fetch the top 3-5 decklists
     # 3. Normalize card names and return as TemplateDeck
-    print(f"    [ARCHIDEKT] Stub: No template decks for '{commander_name}' (not yet implemented)")
+    log.info(f"    Stub: No template decks for '{commander_name}' (not yet implemented)")
     return []

deck_sources/edhrec_adapter.py

@@ -7,6 +7,10 @@ Currently returns a hardcoded sample for end-to-end wiring.
 Real implementation will parse EDHREC commander pages.
 """
 from .models import TemplateDeck, TemplateDeckCard
+import logging
+
+log = logging.getLogger("commander_ai_lab.deckgen.edhrec")
+
 
 
 def fetch_template_decks(commander_name: str, color_identity: list, config: dict = None) -> list:
@@ -27,5 +31,5 @@ def fetch_template_decks(commander_name: str, color_identity: list, config: dict
     # 2. Fetch the "Average Deck" or "Top Cards" list
     # 3. Parse card names and inclusion percentages
     # 4. Return as a single TemplateDeck with weighted cards
-    print(f"    [EDHREC] Stub: No recommendations for '{commander_name}' (not yet implemented)")
+    log.info(f"    Stub: No recommendations for '{commander_name}' (not yet implemented)")
     return []

deck_sources/moxfield_adapter.py

@@ -1,17 +1,133 @@
 """
-Moxfield Adapter (stub)
-════════════════════════
-
+Moxfield Adapter
+================
 Fetches template decklists from Moxfield for a given commander.
-Currently returns empty (experimental — not yet implemented).
+Uses the undocumented but usable api2.moxfield.com API.
+Rate-limited to <=10 req/sec. Errors are caught and logged.
 """
+import logging
+import time
+
+import httpx
+
 from .models import TemplateDeck, TemplateDeckCard
 
+log = logging.getLogger("commander_ai_lab.deckgen.moxfield")
 
-def fetch_template_decks(commander_name: str, color_identity: list, config: dict = None) -> list:
+MOXFIELD_API = "https://api2.moxfield.com"
+HEADERS = {
+    "User-Agent": "CommanderAILab/1.0",
+    "Accept": "application/json",
+}
+# Minimum seconds between requests (<=10 req/sec)
+_RATE_LIMIT_DELAY = 0.12
+
+
+def _slugify_commander(name: str) -> str:
+    """Return the commander name as-is for the Moxfield search query."""
+    return name.strip()
+
+
+def _build_cards_from_deck(deck_data: dict) -> list:
+    """Extract commander + mainboard cards from a full Moxfield deck response."""
+    cards = []
+    # Commanders first
+    for card_name, card_obj in deck_data.get("commanders", {}).items():
+        qty = card_obj.get("quantity", 1)
+        cards.append(TemplateDeckCard(name=card_name, quantity=qty))
+    # Mainboard
+    for card_name, card_obj in deck_data.get("mainboard", {}).items():
+        qty = card_obj.get("quantity", 1)
+        cards.append(TemplateDeckCard(name=card_name, quantity=qty))
+    return cards
+
+
+def fetch_template_decks(
+    commander_name: str,
+    color_identity: list,
+    config: dict = None,
+) -> list:
     """
-    Fetch template decks from Moxfield for the given commander.
-    Experimental — not yet implemented.
+    Fetch top Commander decks from Moxfield for the given commander.
+
+    Returns a list of TemplateDeck objects. Returns empty list on any
+    API failure — errors are logged, never raised.
+
+    Args:
+        commander_name: The commander's card name.
+        color_identity: List of color symbols (unused, kept for interface compat).
+        config: Optional dict. Supported key: ``max_decks`` (default 3).
+
+    Returns:
+        List of TemplateDeck objects populated with commander + mainboard cards.
     """
-    print(f"    [MOXFIELD] Stub: Not yet implemented for '{commander_name}'")
-    return []
+    max_decks = (config or {}).get("max_decks", 3)
+    results = []
+
+    try:
+        # Step 1 — Search for top decks by commander name
+        search_url = f"{MOXFIELD_API}/v2/decks/search"
+        params = {
+            "q": "",
+            "fmt": "commander",
+            "commander": _slugify_commander(commander_name),
+            "sortType": "views",
+            "sortDirection": "Descending",
+            "pageSize": max_decks,
+        }
+        log.info(f" Moxfield: Searching for '{commander_name}' (max={max_decks})")
+        resp = httpx.get(search_url, params=params, headers=HEADERS, timeout=15)
+        resp.raise_for_status()
+        search_data = resp.json()
+
+        deck_summaries = search_data.get("data", [])[:max_decks]
+        if not deck_summaries:
+            log.info(f" Moxfield: No decks found for '{commander_name}'")
+            return []
+
+        log.info(f" Moxfield: Found {len(deck_summaries)} deck(s) for '{commander_name}', fetching details...")
+
+        # Step 2 — Fetch full detail for each deck
+        for summary in deck_summaries:
+            deck_id = summary.get("publicId")
+            deck_name = summary.get("name", f"Moxfield {commander_name}")
+            if not deck_id:
+                log.warning(" Moxfield: Skipping deck with no publicId")
+                continue
+
+            time.sleep(_RATE_LIMIT_DELAY)  # Rate limit: <=10 req/sec
+
+            try:
+                deck_url = f"{MOXFIELD_API}/v3/decks/all/{deck_id}"
+                deck_resp = httpx.get(deck_url, headers=HEADERS, timeout=15)
+                deck_resp.raise_for_status()
+                deck_data = deck_resp.json()
+            except Exception as deck_err:
+                log.warning(f" Moxfield: Failed to fetch deck '{deck_id}': {deck_err}")
+                continue
+
+            cards = _build_cards_from_deck(deck_data)
+            if not cards:
+                log.warning(f" Moxfield: Deck '{deck_name}' ({deck_id}) had no cards, skipping")
+                continue
+
+            results.append(TemplateDeck(
+                name=f"{deck_name} (Moxfield)",
+                source=f"https://moxfield.com/decks/{deck_id}",
+                cards=cards,
+            ))
+            log.info(f" Moxfield: Added deck '{deck_name}' ({len(cards)} cards)")
+
+        log.info(f" Moxfield: Returning {len(results)} deck(s) for '{commander_name}'")
+
+    except httpx.HTTPStatusError as e:
+        log.warning(
+            f" Moxfield fetch failed for '{commander_name}': "
+            f"HTTP {e.response.status_code} — {e}"
+        )
+    except httpx.RequestError as e:
+        log.warning(f" Moxfield request error for '{commander_name}': {e}")
+    except Exception as e:
+        log.warning(f" Moxfield unexpected error for '{commander_name}': {e}")
+
+    return results

deck_sources/mtggoldfish_adapter.py

@@ -6,6 +6,10 @@ Fetches template decklists from MTGGoldfish for a given commander.
 Currently returns empty (experimental — not yet implemented).
 """
 from .models import TemplateDeck, TemplateDeckCard
+import logging
+
+log = logging.getLogger("commander_ai_lab.deckgen.mtggoldfish")
+
 
 
 def fetch_template_decks(commander_name: str, color_identity: list, config: dict = None) -> list:
@@ -13,5 +17,5 @@ def fetch_template_decks(commander_name: str, color_identity: list, config: dict
     Fetch template decks from MTGGoldfish for the given commander.
     Experimental — not yet implemented.
     """
-    print(f"    [MTGGOLDFISH] Stub: Not yet implemented for '{commander_name}'")
+    log.info(f"    Stub: Not yet implemented for '{commander_name}'")
     return []

frontend/.gitignore

@@ -0,0 +1,24 @@
+# Logs
+logs
+*.log
+npm-debug.log*
+yarn-debug.log*
+yarn-error.log*
+pnpm-debug.log*
+lerna-debug.log*
+
+node_modules
+dist
+dist-ssr
+*.local
+
+# Editor directories and files
+.vscode/*
+!.vscode/extensions.json
+.idea
+.DS_Store
+*.suo
+*.ntvs*
+*.njsproj
+*.sln
+*.sw?

frontend/commander-ai-lab-ui/src/pages/SimulatorPage.tsx

@@ -133,7 +133,7 @@ export function SimulatorPage() {
           />
           <input
             type="text"
-            placeholder="Base URL (e.g. http://192.168.0.122:1234)"
+            placeholder="Base URL (e.g. http://192.168.0.240:11434)"
             value={dsUrl}
             onChange={e => setDsUrl(e.target.value)}
             className="w-full px-3 py-2 bg-bg-tertiary border border-border-primary rounded-lg text-sm text-text-primary placeholder:text-text-tertiary focus:outline-none focus:border-accent-blue/50 transition-all"

frontend/commander-ai-lab-ui/src/types/index.ts

@@ -559,5 +559,4 @@ export interface CommanderSearchResult {
 
 export interface PplxStatus {
   configured: boolean
-  key_prefix: string
 }

frontend/eslint.config.js

@@ -0,0 +1,23 @@
+import js from '@eslint/js'
+import globals from 'globals'
+import reactHooks from 'eslint-plugin-react-hooks'
+import reactRefresh from 'eslint-plugin-react-refresh'
+import tseslint from 'typescript-eslint'
+import { defineConfig, globalIgnores } from 'eslint/config'
+
+export default defineConfig([
+  globalIgnores(['dist']),
+  {
+    files: ['**/*.{ts,tsx}'],
+    extends: [
+      js.configs.recommended,
+      tseslint.configs.recommended,
+      reactHooks.configs.flat.recommended,
+      reactRefresh.configs.vite,
+    ],
+    languageOptions: {
+      ecmaVersion: 2020,
+      globals: globals.browser,
+    },
+  },
+])

frontend/index.html

@@ -0,0 +1,13 @@
+<!doctype html>
+<html lang="en">
+  <head>
+    <meta charset="UTF-8" />
+    <link rel="icon" type="image/svg+xml" href="/favicon.svg" />
+    <meta name="viewport" content="width=device-width, initial-scale=1.0" />
+    <title>frontend-tmp</title>
+  </head>
+  <body>
+    <div id="root"></div>
+    <script type="module" src="/src/main.tsx"></script>
+  </body>
+</html>

frontend/package.json

@@ -0,0 +1,42 @@
+{
+  "name": "commander-ai-lab-frontend",
+  "private": true,
+  "version": "0.0.0",
+  "type": "module",
+  "scripts": {
+    "dev": "vite",
+    "build": "tsc -b && vite build",
+    "lint": "eslint .",
+    "preview": "vite preview",
+    "test": "vitest run",
+    "test:watch": "vitest"
+  },
+  "dependencies": {
+    "clsx": "^2.1.1",
+    "react": "^19.2.4",
+    "react-dom": "^19.2.4",
+    "react-router-dom": "^7.13.1"
+  },
+  "devDependencies": {
+    "@eslint/js": "^9.39.4",
+    "@testing-library/jest-dom": "^6.9.1",
+    "@testing-library/react": "^16.3.2",
+    "@testing-library/user-event": "^14.6.1",
+    "@types/node": "^24.12.0",
+    "@types/react": "^19.2.14",
+    "@types/react-dom": "^19.2.3",
+    "@vitejs/plugin-react": "^6.0.0",
+    "autoprefixer": "^10.4.27",
+    "eslint": "^9.39.4",
+    "eslint-plugin-react-hooks": "^7.0.1",
+    "eslint-plugin-react-refresh": "^0.5.2",
+    "globals": "^17.4.0",
+    "jsdom": "^29.0.0",
+    "postcss": "^8.5.8",
+    "tailwindcss": "^3.4.19",
+    "typescript": "~5.9.3",
+    "typescript-eslint": "^8.56.1",
+    "vite": "^8.0.0",
+    "vitest": "^4.1.0"
+  }
+}

frontend/postcss.config.js

@@ -0,0 +1,6 @@
+export default {
+  plugins: {
+    tailwindcss: {},
+    autoprefixer: {},
+  },
+}

frontend/public/favicon.svg

@@ -0,0 +1 @@
+<svg xmlns="http://www.w3.org/2000/svg" width="48" height="46" fill="none" viewBox="0 0 48 46"><path fill="#863bff" d="M25.946 44.938c-.664.845-2.021.375-2.021-.698V33.937a2.26 2.26 0 0 0-2.262-2.262H10.287c-.92 0-1.456-1.04-.92-1.788l7.48-10.471c1.07-1.497 0-3.578-1.842-3.578H1.237c-.92 0-1.456-1.04-.92-1.788L10.013.474c.214-.297.556-.474.92-.474h28.894c.92 0 1.456 1.04.92 1.788l-7.48 10.471c-1.07 1.498 0 3.579 1.842 3.579h11.377c.943 0 1.473 1.088.89 1.83L25.947 44.94z" style="fill:#863bff;fill:color(display-p3 .5252 .23 1);fill-opacity:1"/><mask id="a" width="48" height="46" x="0" y="0" maskUnits="userSpaceOnUse" style="mask-type:alpha"><path fill="#000" d="M25.842 44.938c-.664.844-2.021.375-2.021-.698V33.937a2.26 2.26 0 0 0-2.262-2.262H10.183c-.92 0-1.456-1.04-.92-1.788l7.48-10.471c1.07-1.498 0-3.579-1.842-3.579H1.133c-.92 0-1.456-1.04-.92-1.787L9.91.473c.214-.297.556-.474.92-.474h28.894c.92 0 1.456 1.04.92 1.788l-7.48 10.471c-1.07 1.498 0 3.578 1.842 3.578h11.377c.943 0 1.473 1.088.89 1.832L25.843 44.94z" style="fill:#000;fill-opacity:1"/></mask><g mask="url(#a)"><g filter="url(#b)"><ellipse cx="5.508" cy="14.704" fill="#ede6ff" rx="5.508" ry="14.704" style="fill:#ede6ff;fill:color(display-p3 .9275 .9033 1);fill-opacity:1" transform="matrix(.00324 1 1 -.00324 -4.47 31.516)"/></g><g filter="url(#c)"><ellipse cx="10.399" cy="29.851" fill="#ede6ff" rx="10.399" ry="29.851" style="fill:#ede6ff;fill:color(display-p3 .9275 .9033 1);fill-opacity:1" transform="matrix(.00324 1 1 -.00324 -39.328 7.883)"/></g><g filter="url(#d)"><ellipse cx="5.508" cy="30.487" fill="#7e14ff" rx="5.508" ry="30.487" style="fill:#7e14ff;fill:color(display-p3 .4922 .0767 1);fill-opacity:1" transform="rotate(89.814 -25.913 -14.639)scale(1 -1)"/></g><g filter="url(#e)"><ellipse cx="5.508" cy="30.599" fill="#7e14ff" rx="5.508" ry="30.599" style="fill:#7e14ff;fill:color(display-p3 .4922 .0767 1);fill-opacity:1" transform="rotate(89.814 -32.644 -3.334)scale(1 -1)"/></g><g filter="url(#f)"><ellipse cx="5.508" cy="30.599" fill="#7e14ff" rx="5.508" ry="30.599" style="fill:#7e14ff;fill:color(display-p3 .4922 .0767 1);fill-opacity:1" transform="matrix(.00324 1 1 -.00324 -34.34 30.47)"/></g><g filter="url(#g)"><ellipse cx="14.072" cy="22.078" fill="#ede6ff" rx="14.072" ry="22.078" style="fill:#ede6ff;fill:color(display-p3 .9275 .9033 1);fill-opacity:1" transform="rotate(93.35 24.506 48.493)scale(-1 1)"/></g><g filter="url(#h)"><ellipse cx="3.47" cy="21.501" fill="#7e14ff" rx="3.47" ry="21.501" style="fill:#7e14ff;fill:color(display-p3 .4922 .0767 1);fill-opacity:1" transform="rotate(89.009 28.708 47.59)scale(-1 1)"/></g><g filter="url(#i)"><ellipse cx="3.47" cy="21.501" fill="#7e14ff" rx="3.47" ry="21.501" style="fill:#7e14ff;fill:color(display-p3 .4922 .0767 1);fill-opacity:1" transform="rotate(89.009 28.708 47.59)scale(-1 1)"/></g><g filter="url(#j)"><ellipse cx=".387" cy="8.972" fill="#7e14ff" rx="4.407" ry="29.108" style="fill:#7e14ff;fill:color(display-p3 .4922 .0767 1);fill-opacity:1" transform="rotate(39.51 .387 8.972)"/></g><g filter="url(#k)"><ellipse cx="47.523" cy="-6.092" fill="#7e14ff" rx="4.407" ry="29.108" style="fill:#7e14ff;fill:color(display-p3 .4922 .0767 1);fill-opacity:1" transform="rotate(37.892 47.523 -6.092)"/></g><g filter="url(#l)"><ellipse cx="41.412" cy="6.333" fill="#47bfff" rx="5.971" ry="9.665" style="fill:#47bfff;fill:color(display-p3 .2799 .748 1);fill-opacity:1" transform="rotate(37.892 41.412 6.333)"/></g><g filter="url(#m)"><ellipse cx="-1.879" cy="38.332" fill="#7e14ff" rx="4.407" ry="29.108" style="fill:#7e14ff;fill:color(display-p3 .4922 .0767 1);fill-opacity:1" transform="rotate(37.892 -1.88 38.332)"/></g><g filter="url(#n)"><ellipse cx="-1.879" cy="38.332" fill="#7e14ff" rx="4.407" ry="29.108" style="fill:#7e14ff;fill:color(display-p3 .4922 .0767 1);fill-opacity:1" transform="rotate(37.892 -1.88 38.332)"/></g><g filter="url(#o)"><ellipse cx="35.651" cy="29.907" fill="#7e14ff" rx="4.407" ry="29.108" style="fill:#7e14ff;fill:color(display-p3 .4922 .0767 1);fill-opacity:1" transform="rotate(37.892 35.651 29.907)"/></g><g filter="url(#p)"><ellipse cx="38.418" cy="32.4" fill="#47bfff" rx="5.971" ry="15.297" style="fill:#47bfff;fill:color(display-p3 .2799 .748 1);fill-opacity:1" transform="rotate(37.892 38.418 32.4)"/></g></g><defs><filter id="b" width="60.045" height="41.654" x="-19.77" y="16.149" color-interpolation-filters="sRGB" filterUnits="userSpaceOnUse"><feFlood flood-opacity="0" result="BackgroundImageFix"/><feBlend in="SourceGraphic" in2="BackgroundImageFix" result="shape"/><feGaussianBlur result="effect1_foregroundBlur_2002_17158" stdDeviation="7.659"/></filter><filter id="c" width="90.34" height="51.437" x="-54.613" y="-7.533" color-interpolation-filters="sRGB" filterUnits="userSpaceOnUse"><feFlood flood-opacity="0" result="BackgroundImageFix"/><feBlend in="SourceGraphic" in2="BackgroundImageFix" result="shape"/><feGaussianBlur result="effect1_foregroundBlur_2002_17158" stdDeviation="7.659"/></filtLine truncated

frontend/public/icons.svg

@@ -0,0 +1,24 @@
+<svg xmlns="http://www.w3.org/2000/svg">
+  <symbol id="bluesky-icon" viewBox="0 0 16 17">
+    <g clip-path="url(#bluesky-clip)"><path fill="#08060d" d="M7.75 7.735c-.693-1.348-2.58-3.86-4.334-5.097-1.68-1.187-2.32-.981-2.74-.79C.188 2.065.1 2.812.1 3.251s.241 3.602.398 4.13c.52 1.744 2.367 2.333 4.07 2.145-2.495.37-4.71 1.278-1.805 4.512 3.196 3.309 4.38-.71 4.987-2.746.608 2.036 1.307 5.91 4.93 2.746 2.72-2.746.747-4.143-1.747-4.512 1.702.189 3.55-.4 4.07-2.145.156-.528.397-3.691.397-4.13s-.088-1.186-.575-1.406c-.42-.19-1.06-.395-2.741.79-1.755 1.24-3.64 3.752-4.334 5.099"/></g>
+    <defs><clipPath id="bluesky-clip"><path fill="#fff" d="M.1.85h15.3v15.3H.1z"/></clipPath></defs>
+  </symbol>
+  <symbol id="discord-icon" viewBox="0 0 20 19">
+    <path fill="#08060d" d="M16.224 3.768a14.5 14.5 0 0 0-3.67-1.153c-.158.286-.343.67-.47.976a13.5 13.5 0 0 0-4.067 0c-.128-.306-.317-.69-.476-.976A14.4 14.4 0 0 0 3.868 3.77C1.546 7.28.916 10.703 1.231 14.077a14.7 14.7 0 0 0 4.5 2.306q.545-.748.965-1.587a9.5 9.5 0 0 1-1.518-.74q.191-.14.372-.293c2.927 1.369 6.107 1.369 8.999 0q.183.152.372.294-.723.437-1.52.74.418.838.963 1.588a14.6 14.6 0 0 0 4.504-2.308c.37-3.911-.63-7.302-2.644-10.309m-9.13 8.234c-.878 0-1.599-.82-1.599-1.82 0-.998.705-1.82 1.6-1.82.894 0 1.614.82 1.599 1.82.001 1-.705 1.82-1.6 1.82m5.91 0c-.878 0-1.599-.82-1.599-1.82 0-.998.705-1.82 1.6-1.82.893 0 1.614.82 1.599 1.82 0 1-.706 1.82-1.6 1.82"/>
+  </symbol>
+  <symbol id="documentation-icon" viewBox="0 0 21 20">
+    <path fill="none" stroke="#aa3bff" stroke-linecap="round" stroke-linejoin="round" stroke-width="1.35" d="m15.5 13.333 1.533 1.322c.645.555.967.833.967 1.178s-.322.623-.967 1.179L15.5 18.333m-3.333-5-1.534 1.322c-.644.555-.966.833-.966 1.178s.322.623.966 1.179l1.534 1.321"/>
+    <path fill="none" stroke="#aa3bff" stroke-linecap="round" stroke-linejoin="round" stroke-width="1.35" d="M17.167 10.836v-4.32c0-1.41 0-2.117-.224-2.68-.359-.906-1.118-1.621-2.08-1.96-.599-.21-1.349-.21-2.848-.21-2.623 0-3.935 0-4.983.369-1.684.591-3.013 1.842-3.641 3.428C3 6.449 3 7.684 3 10.154v2.122c0 2.558 0 3.838.706 4.726q.306.383.713.671c.76.536 1.79.64 3.581.66"/>
+    <path fill="none" stroke="#aa3bff" stroke-linecap="round" stroke-linejoin="round" stroke-width="1.35" d="M3 10a2.78 2.78 0 0 1 2.778-2.778c.555 0 1.209.097 1.748-.047.48-.129.854-.503.982-.982.145-.54.048-1.194.048-1.749a2.78 2.78 0 0 1 2.777-2.777"/>
+  </symbol>
+  <symbol id="github-icon" viewBox="0 0 19 19">
+    <path fill="#08060d" fill-rule="evenodd" d="M9.356 1.85C5.05 1.85 1.57 5.356 1.57 9.694a7.84 7.84 0 0 0 5.324 7.44c.387.079.528-.168.528-.376 0-.182-.013-.805-.013-1.454-2.165.467-2.616-.935-2.616-.935-.349-.91-.864-1.143-.864-1.143-.71-.48.051-.48.051-.48.787.051 1.2.805 1.2.805.695 1.194 1.817.857 2.268.649.064-.507.27-.857.49-1.052-1.728-.182-3.545-.857-3.545-3.87 0-.857.31-1.558.8-2.104-.078-.195-.349-1 .077-2.078 0 0 .657-.208 2.14.805a7.5 7.5 0 0 1 1.946-.26c.657 0 1.328.092 1.946.26 1.483-1.013 2.14-.805 2.14-.805.426 1.078.155 1.883.078 2.078.502.546.799 1.247.799 2.104 0 3.013-1.818 3.675-3.558 3.87.284.247.528.714.528 1.454 0 1.052-.012 1.896-.012 2.156 0 .208.142.455.528.377a7.84 7.84 0 0 0 5.324-7.441c.013-4.338-3.48-7.844-7.773-7.844" clip-rule="evenodd"/>
+  </symbol>
+  <symbol id="social-icon" viewBox="0 0 20 20">
+    <path fill="none" stroke="#aa3bff" stroke-linecap="round" stroke-linejoin="round" stroke-width="1.35" d="M12.5 6.667a4.167 4.167 0 1 0-8.334 0 4.167 4.167 0 0 0 8.334 0"/>
+    <path fill="none" stroke="#aa3bff" stroke-linecap="round" stroke-linejoin="round" stroke-width="1.35" d="M2.5 16.667a5.833 5.833 0 0 1 8.75-5.053m3.837.474.513 1.035c.07.144.257.282.414.309l.93.155c.596.1.736.536.307.965l-.723.73a.64.64 0 0 0-.152.531l.207.903c.164.715-.213.991-.84.618l-.872-.52a.63.63 0 0 0-.577 0l-.872.52c-.624.373-1.003.094-.84-.618l.207-.903a.64.64 0 0 0-.152-.532l-.723-.729c-.426-.43-.289-.864.306-.964l.93-.156a.64.64 0 0 0 .412-.31l.513-1.034c.28-.562.735-.562 1.012 0"/>
+  </symbol>
+  <symbol id="x-icon" viewBox="0 0 19 19">
+    <path fill="#08060d" fill-rule="evenodd" d="M1.893 1.98c.052.072 1.245 1.769 2.653 3.77l2.892 4.114c.183.261.333.48.333.486s-.068.089-.152.183l-.522.593-.765.867-3.597 4.087c-.375.426-.734.834-.798.905a1 1 0 0 0-.118.148c0 .01.236.017.664.017h.663l.729-.83c.4-.457.796-.906.879-.999a692 692 0 0 0 1.794-2.038c.034-.037.301-.34.594-.675l.551-.624.345-.392a7 7 0 0 1 .34-.374c.006 0 .93 1.306 2.052 2.903l2.084 2.965.045.063h2.275c1.87 0 2.273-.003 2.266-.021-.008-.02-1.098-1.572-3.894-5.547-2.013-2.862-2.28-3.246-2.273-3.266.008-.019.282-.332 2.085-2.38l2-2.274 1.567-1.782c.022-.028-.016-.03-.65-.03h-.674l-.3.342a871 871 0 0 1-1.782 2.025c-.067.075-.405.458-.75.852a100 100 0 0 1-.803.91c-.148.172-.299.344-.99 1.127-.304.343-.32.358-.345.327-.015-.019-.904-1.282-1.976-2.808L6.365 1.85H1.8zm1.782.91 8.078 11.294c.772 1.08 1.413 1.973 1.425 1.984.016.017.241.02 1.05.017l1.03-.004-2.694-3.766L7.796 5.75 5.722 2.852l-1.039-.004-1.039-.004z" clip-rule="evenodd"/>
+  </symbol>
+</svg>

frontend/src/App.css

@@ -0,0 +1,184 @@
+.counter {
+  font-size: 16px;
+  padding: 5px 10px;
+  border-radius: 5px;
+  color: var(--accent);
+  background: var(--accent-bg);
+  border: 2px solid transparent;
+  transition: border-color 0.3s;
+  margin-bottom: 24px;
+
+  &:hover {
+    border-color: var(--accent-border);
+  }
+  &:focus-visible {
+    outline: 2px solid var(--accent);
+    outline-offset: 2px;
+  }
+}
+
+.hero {
+  position: relative;
+
+  .base,
+  .framework,
+  .vite {
+    inset-inline: 0;
+    margin: 0 auto;
+  }
+
+  .base {
+    width: 170px;
+    position: relative;
+    z-index: 0;
+  }
+
+  .framework,
+  .vite {
+    position: absolute;
+  }
+
+  .framework {
+    z-index: 1;
+    top: 34px;
+    height: 28px;
+    transform: perspective(2000px) rotateZ(300deg) rotateX(44deg) rotateY(39deg)
+      scale(1.4);
+  }
+
+  .vite {
+    z-index: 0;
+    top: 107px;
+    height: 26px;
+    width: auto;
+    transform: perspective(2000px) rotateZ(300deg) rotateX(40deg) rotateY(39deg)
+      scale(0.8);
+  }
+}
+
+#center {
+  display: flex;
+  flex-direction: column;
+  gap: 25px;
+  place-content: center;
+  place-items: center;
+  flex-grow: 1;
+
+  @media (max-width: 1024px) {
+    padding: 32px 20px 24px;
+    gap: 18px;
+  }
+}
+
+#next-steps {
+  display: flex;
+  border-top: 1px solid var(--border);
+  text-align: left;
+
+  & > div {
+    flex: 1 1 0;
+    padding: 32px;
+    @media (max-width: 1024px) {
+      padding: 24px 20px;
+    }
+  }
+
+  .icon {
+    margin-bottom: 16px;
+    width: 22px;
+    height: 22px;
+  }
+
+  @media (max-width: 1024px) {
+    flex-direction: column;
+    text-align: center;
+  }
+}
+
+#docs {
+  border-right: 1px solid var(--border);
+
+  @media (max-width: 1024px) {
+    border-right: none;
+    border-bottom: 1px solid var(--border);
+  }
+}
+
+#next-steps ul {
+  list-style: none;
+  padding: 0;
+  display: flex;
+  gap: 8px;
+  margin: 32px 0 0;
+
+  .logo {
+    height: 18px;
+  }
+
+  a {
+    color: var(--text-h);
+    font-size: 16px;
+    border-radius: 6px;
+    background: var(--social-bg);
+    display: flex;
+    padding: 6px 12px;
+    align-items: center;
+    gap: 8px;
+    text-decoration: none;
+    transition: box-shadow 0.3s;
+
+    &:hover {
+      box-shadow: var(--shadow);
+    }
+    .button-icon {
+      height: 18px;
+      width: 18px;
+    }
+  }
+
+  @media (max-width: 1024px) {
+    margin-top: 20px;
+    flex-wrap: wrap;
+    justify-content: center;
+
+    li {
+      flex: 1 1 calc(50% - 8px);
+    }
+
+    a {
+      width: 100%;
+      justify-content: center;
+      box-sizing: border-box;
+    }
+  }
+}
+
+#spacer {
+  height: 88px;
+  border-top: 1px solid var(--border);
+  @media (max-width: 1024px) {
+    height: 48px;
+  }
+}
+
+.ticks {
+  position: relative;
+  width: 100%;
+
+  &::before,
+  &::after {
+    content: '';
+    position: absolute;
+    top: -4.5px;
+    border: 5px solid transparent;
+  }
+
+  &::before {
+    left: 0;
+    border-left-color: var(--border);
+  }
+  &::after {
+    right: 0;
+    border-right-color: var(--border);
+  }
+}

frontend/src/App.tsx

@@ -0,0 +1,48 @@
+import { BrowserRouter, Routes, Route, Link, useLocation } from 'react-router-dom';
+import LobbyPage from './pages/LobbyPage';
+import RoomPage from './pages/RoomPage';
+import GamePage from './pages/GamePage';
+import SimulatorPage from './pages/SimulatorPage';
+
+function Nav() {
+  const location = useLocation();
+  // Hide nav on active game pages
+  if (location.pathname.startsWith('/game/')) return null;
+  const links = [
+    { to: '/', label: 'Lobby' },
+    { to: '/simulator', label: 'Simulator' },
+  ];
+  return (
+    <nav className="bg-bg-bar border-b border-border-zone px-4 py-2 flex items-center gap-4">
+      <span className="text-sm font-bold text-white mr-4">Commander AI Lab</span>
+      {links.map((l) => (
+        <Link
+          key={l.to}
+          to={l.to}
+          className={`text-sm px-2 py-1 rounded transition-colors ${
+            location.pathname === l.to
+              ? 'text-white bg-accent-blue/20'
+              : 'text-text-muted hover:text-white'
+          }`}
+        >
+          {l.label}
+        </Link>
+      ))}
+    </nav>
+  );
+}
+
+function App() {
+  return (
+    <BrowserRouter>
+      <Nav />
+      <Routes>
+        <Route path="/" element={<LobbyPage />} />
+        <Route path="/room/:roomId" element={<RoomPage />} />
+        <Route path="/game/:gameId" element={<GamePage />} />
+        <Route path="/simulator" element={<SimulatorPage />} />
+      </Routes>
+    </BrowserRouter>
+  );
+}
+export default App;

frontend/src/__tests__/CardChip.test.tsx

@@ -0,0 +1,35 @@
+import { render, screen } from '@testing-library/react';
+import { describe, it, expect } from 'vitest';
+import CardChip from '../components/CardChip';
+import type { Card } from '../types/game';
+
+function makeCard(overrides: Partial<Card> = {}): Card {
+  return {
+    name: 'Sol Ring',
+    type: 'Artifact',
+    tapped: false,
+    is_commander: false,
+    oracle: 'Tap: Add CC.',
+    cmc: 1,
+    ...overrides,
+  };
+}
+
+describe('CardChip', () => {
+  it('applies opacity class when tapped', () => {
+    render(<CardChip card={makeCard({ tapped: true })} />);
+    const chip = screen.getByText(/Sol Ring/);
+    expect(chip.closest('span')).toHaveClass('opacity-55');
+  });
+
+  it('shows star for commander cards', () => {
+    render(<CardChip card={makeCard({ is_commander: true })} />);
+    expect(screen.getByText('★')).toBeInTheDocument();
+  });
+
+  it('has hover border class', () => {
+    render(<CardChip card={makeCard()} />);
+    const chip = screen.getByText(/Sol Ring/).closest('span');
+    expect(chip).toHaveClass('hover:border-[#ffdc32]');
+  });
+});

frontend/src/__tests__/GamePage.test.tsx

@@ -0,0 +1,86 @@
+import { render, screen } from '@testing-library/react';
+import { describe, it, expect, vi, beforeEach } from 'vitest';
+import { MemoryRouter, Route, Routes } from 'react-router-dom';
+import GamePage from '../pages/GamePage';
+import type { GameState, Card } from '../types/game';
+
+const mockGameState: GameState = {
+  game_id: 'test-game',
+  turn: 1,
+  current_phase: 'main1',
+  active_seat: 0,
+  players: [
+    {
+      seat: 0, name: 'Alice', life: 40, eliminated: false,
+      hand: [], battlefield: [], graveyard: [], exile: [], command_zone: [], hand_count: 7,
+    },
+    {
+      seat: 1, name: 'Bob', life: 38, eliminated: false,
+      hand: [], battlefield: [], graveyard: [], exile: [], command_zone: [], hand_count: 6,
+    },
+    {
+      seat: 2, name: 'Charlie', life: 40, eliminated: false,
+      hand: [], battlefield: [], graveyard: [], exile: [], command_zone: [], hand_count: 7,
+    },
+    {
+      seat: 3, name: 'Diana', life: 35, eliminated: false,
+      hand: [], battlefield: [], graveyard: [], exile: [], command_zone: [], hand_count: 5,
+    },
+  ],
+  stack: [],
+  game_over: false,
+  winner: null,
+};
+
+const mockHand: Card[] = [];
+const mockSendMove = vi.fn();
+const mockSendConcede = vi.fn();
+
+vi.mock('../hooks/useGameSocket', () => ({
+  useGameSocket: () => ({
+    gameState: mockGameState,
+    narration: [],
+    thinking: new Set<number>(),
+    hand: mockHand,
+    gameOver: null,
+    status: 'Connected',
+    sendMove: mockSendMove,
+    sendConcede: mockSendConcede,
+  }),
+}));
+
+describe('GamePage integration', () => {
+  beforeEach(() => {
+    vi.clearAllMocks();
+  });
+
+  it('renders all 4 PlayerZone components', () => {
+    render(
+      <MemoryRouter initialEntries={['/game/test-game']}>
+        <Routes>
+          <Route path="/game/:gameId" element={<GamePage />} />
+        </Routes>
+      </MemoryRouter>,
+    );
+
+    // P0: Alice appears in both PhaseBar and PlayerZone, so use getAllByText
+    expect(screen.getAllByText(/P0: Alice/).length).toBeGreaterThanOrEqual(1);
+    expect(screen.getAllByText(/P1: Bob/).length).toBeGreaterThanOrEqual(1);
+    expect(screen.getAllByText(/P2: Charlie/).length).toBeGreaterThanOrEqual(1);
+    expect(screen.getAllByText(/P3: Diana/).length).toBeGreaterThanOrEqual(1);
+  });
+
+  it('renders PhaseBar with current phase', () => {
+    render(
+      <MemoryRouter initialEntries={['/game/test-game']}>
+        <Routes>
+          <Route path="/game/:gameId" element={<GamePage />} />
+        </Routes>
+      </MemoryRouter>,
+    );
+
+    // main1 should be highlighted
+    const main1Buttons = screen.getAllByText('main1');
+    expect(main1Buttons.length).toBeGreaterThan(0);
+  });
+});

frontend/src/__tests__/MoveBar.test.tsx

@@ -0,0 +1,66 @@
+import { render, screen } from '@testing-library/react';
+import userEvent from '@testing-library/user-event';
+import { describe, it, expect, vi } from 'vitest';
+import MoveBar from '../components/MoveBar';
+
+describe('MoveBar', () => {
+  it('calls onPass when pass button is clicked', async () => {
+    const onPass = vi.fn();
+    render(
+      <MoveBar
+        hand={[]}
+        humanSeat={0}
+        onPass={onPass}
+        onConcede={vi.fn()}
+        onPlayCard={vi.fn()}
+        status=""
+      />,
+    );
+
+    await userEvent.click(screen.getByText('Pass Priority'));
+    expect(onPass).toHaveBeenCalledOnce();
+  });
+
+  it('shows confirm dialog on concede', async () => {
+    const onConcede = vi.fn();
+    const confirmSpy = vi.spyOn(window, 'confirm').mockReturnValue(true);
+
+    render(
+      <MoveBar
+        hand={[]}
+        humanSeat={0}
+        onPass={vi.fn()}
+        onConcede={onConcede}
+        onPlayCard={vi.fn()}
+        status=""
+      />,
+    );
+
+    await userEvent.click(screen.getByText('Concede'));
+    expect(confirmSpy).toHaveBeenCalledWith('Are you sure you want to concede?');
+    expect(onConcede).toHaveBeenCalledOnce();
+
+    confirmSpy.mockRestore();
+  });
+
+  it('does not concede if confirm is cancelled', async () => {
+    const onConcede = vi.fn();
+    const confirmSpy = vi.spyOn(window, 'confirm').mockReturnValue(false);
+
+    render(
+      <MoveBar
+        hand={[]}
+        humanSeat={0}
+        onPass={vi.fn()}
+        onConcede={onConcede}
+        onPlayCard={vi.fn()}
+        status=""
+      />,
+    );
+
+    await userEvent.click(screen.getByText('Concede'));
+    expect(onConcede).not.toHaveBeenCalled();
+
+    confirmSpy.mockRestore();
+  });
+});

frontend/src/__tests__/NarrationPanel.test.tsx

@@ -0,0 +1,38 @@
+import { render, screen } from '@testing-library/react';
+import { describe, it, expect } from 'vitest';
+import NarrationPanel from '../components/NarrationPanel';
+import type { NarrationLine } from '../types/game';
+
+describe('NarrationPanel', () => {
+  it('renders only 120 lines when given more', () => {
+    const lines: NarrationLine[] = Array.from({ length: 150 }, (_, i) => ({
+      text: `Line ${i}`,
+      isSystem: false,
+      timestamp: Date.now(),
+    }));
+
+    render(<NarrationPanel lines={lines} thinkingSeats={new Set()} />);
+
+    // Should not render the first 30 lines (0-29)
+    expect(screen.queryByText('Line 0')).not.toBeInTheDocument();
+    expect(screen.queryByText('Line 29')).not.toBeInTheDocument();
+
+    // Should render lines 30-149
+    expect(screen.getByText('Line 30')).toBeInTheDocument();
+    expect(screen.getByText('Line 149')).toBeInTheDocument();
+  });
+
+  it('has a scroll container ref', () => {
+    const lines: NarrationLine[] = [
+      { text: 'Hello world', isSystem: false, timestamp: Date.now() },
+    ];
+
+    const { container } = render(
+      <NarrationPanel lines={lines} thinkingSeats={new Set()} />,
+    );
+
+    // The container div has overflow-y-auto
+    const panel = container.firstChild as HTMLElement;
+    expect(panel).toHaveClass('overflow-y-auto');
+  });
+});

frontend/src/__tests__/PhaseBar.test.tsx

@@ -0,0 +1,38 @@
+import { render, screen } from '@testing-library/react';
+import { describe, it, expect } from 'vitest';
+import PhaseBar from '../components/PhaseBar';
+import type { Player } from '../types/game';
+
+const players: Player[] = [
+  {
+    seat: 0, name: 'Alice', life: 40, eliminated: false,
+    hand: [], battlefield: [], graveyard: [], exile: [], command_zone: [], hand_count: 7,
+  },
+  {
+    seat: 1, name: 'Bob', life: 40, eliminated: false,
+    hand: [], battlefield: [], graveyard: [], exile: [], command_zone: [], hand_count: 7,
+  },
+];
+
+describe('PhaseBar', () => {
+  it('highlights the active phase tab', () => {
+    render(<PhaseBar phase="main1" activeSeat={0} players={players} />);
+
+    const main1 = screen.getByText('main1');
+    expect(main1).toHaveClass('bg-accent-blue');
+  });
+
+  it('does not highlight inactive phase tabs', () => {
+    render(<PhaseBar phase="main1" activeSeat={0} players={players} />);
+
+    const untap = screen.getByText('untap');
+    expect(untap).not.toHaveClass('bg-accent-blue');
+    expect(untap).toHaveClass('bg-[#252a42]');
+  });
+
+  it('shows active player name', () => {
+    render(<PhaseBar phase="main1" activeSeat={0} players={players} />);
+
+    expect(screen.getByText(/P0: Alice/)).toBeInTheDocument();
+  });
+});

frontend/src/__tests__/PlayerZone.test.tsx

@@ -0,0 +1,32 @@
+import { render, screen } from '@testing-library/react';
+import { describe, it, expect } from 'vitest';
+import PlayerZone from '../components/PlayerZone';
+import type { Player } from '../types/game';
+
+function makePlayer(overrides: Partial<Player> = {}): Player {
+  return {
+    seat: 0, name: 'Alice', life: 40, eliminated: false,
+    hand: [], battlefield: [], graveyard: [], exile: [], command_zone: [], hand_count: 7,
+    ...overrides,
+  };
+}
+
+describe('PlayerZone life colors', () => {
+  it('uses life-high color for life >= 30', () => {
+    render(<PlayerZone player={makePlayer({ life: 35 })} isActive={false} isThinking={false} seatColour="#4696ff" />);
+    const life = screen.getByText('35');
+    expect(life).toHaveClass('text-life-high');
+  });
+
+  it('uses life-med color for life >= 15 and < 30', () => {
+    render(<PlayerZone player={makePlayer({ life: 20 })} isActive={false} isThinking={false} seatColour="#4696ff" />);
+    const life = screen.getByText('20');
+    expect(life).toHaveClass('text-life-med');
+  });
+
+  it('uses life-low color for life < 15', () => {
+    render(<PlayerZone player={makePlayer({ life: 10 })} isActive={false} isThinking={false} seatColour="#4696ff" />);
+    const life = screen.getByText('10');
+    expect(life).toHaveClass('text-life-low');
+  });
+});

frontend/src/__tests__/gameReducer.test.ts

@@ -0,0 +1,78 @@
+import { describe, it, expect } from 'vitest';
+import { gameReducer, initialState } from '../store/gameReducer';
+import type { GameState, Card, NarrationLine } from '../types/game';
+
+const mockGameState: GameState = {
+  game_id: 'g1',
+  turn: 1,
+  current_phase: 'main1',
+  active_seat: 0,
+  players: [],
+  stack: [],
+  game_over: false,
+  winner: null,
+};
+
+const mockCard: Card = {
+  name: 'Sol Ring',
+  type: 'Artifact',
+  tapped: false,
+  is_commander: false,
+  oracle: '',
+  cmc: 1,
+};
+
+const mockLine: NarrationLine = {
+  text: 'Test narration',
+  isSystem: false,
+  timestamp: Date.now(),
+};
+
+describe('gameReducer', () => {
+  it('SET_STATE sets the game state', () => {
+    const result = gameReducer(initialState, { type: 'SET_STATE', payload: mockGameState });
+    expect(result.gameState).toEqual(mockGameState);
+  });
+
+  it('SET_HAND sets the hand cards', () => {
+    const cards = [mockCard];
+    const result = gameReducer(initialState, { type: 'SET_HAND', cards });
+    expect(result.hand).toEqual(cards);
+  });
+
+  it('ADD_NARRATION appends a line', () => {
+    const result = gameReducer(initialState, { type: 'ADD_NARRATION', line: mockLine });
+    expect(result.narration).toHaveLength(1);
+    expect(result.narration[0].text).toBe('Test narration');
+  });
+
+  it('ADD_NARRATION caps at 120 lines', () => {
+    let state = initialState;
+    for (let i = 0; i < 130; i++) {
+      state = gameReducer(state, {
+        type: 'ADD_NARRATION',
+        line: { text: `line ${i}`, isSystem: false, timestamp: Date.now() },
+      });
+    }
+    expect(state.narration).toHaveLength(120);
+    expect(state.narration[0].text).toBe('line 10');
+  });
+
+  it('SET_THINKING adds and removes seats', () => {
+    let state = gameReducer(initialState, { type: 'SET_THINKING', seat: 2, active: true });
+    expect(state.thinkingSeats.has(2)).toBe(true);
+
+    state = gameReducer(state, { type: 'SET_THINKING', seat: 2, active: false });
+    expect(state.thinkingSeats.has(2)).toBe(false);
+  });
+
+  it('GAME_OVER sets winner and reason', () => {
+    const result = gameReducer(initialState, { type: 'GAME_OVER', winner: 1, reason: 'elimination' });
+    expect(result.gameOver).toEqual({ winner: 1, reason: 'elimination' });
+  });
+
+  it('SET_STATUS sets status text', () => {
+    const result = gameReducer(initialState, { type: 'SET_STATUS', status: 'Connected' });
+    expect(result.status).toBe('Connected');
+  });
+});

frontend/src/__tests__/useGameSocket.test.ts

@@ -0,0 +1,92 @@
+import { describe, it, expect, vi, beforeEach, afterEach } from 'vitest';
+import { renderHook, act } from '@testing-library/react';
+import { useGameSocket } from '../hooks/useGameSocket';
+
+// Mock WebSocket
+class MockWebSocket {
+  static readonly CONNECTING = 0;
+  static readonly OPEN = 1;
+  static readonly CLOSING = 2;
+  static readonly CLOSED = 3;
+
+  static instances: MockWebSocket[] = [];
+  url: string;
+  readyState = 1; // OPEN
+  onopen: ((ev: Event) => void) | null = null;
+  onmessage: ((ev: MessageEvent) => void) | null = null;
+  onclose: ((ev: CloseEvent) => void) | null = null;
+  onerror: ((ev: Event) => void) | null = null;
+  sent: string[] = [];
+
+  constructor(url: string) {
+    this.url = url;
+    MockWebSocket.instances.push(this);
+    // Simulate connection open
+    setTimeout(() => this.onopen?.(new Event('open')), 0);
+  }
+
+  send(data: string) {
+    this.sent.push(data);
+  }
+
+  close() {
+    this.readyState = 3;
+  }
+
+  simulateMessage(data: unknown) {
+    this.onmessage?.(new MessageEvent('message', { data: JSON.stringify(data) }));
+  }
+}
+
+describe('useGameSocket', () => {
+  beforeEach(() => {
+    MockWebSocket.instances = [];
+    vi.stubGlobal('WebSocket', MockWebSocket);
+    vi.useFakeTimers();
+  });
+
+  afterEach(() => {
+    vi.useRealTimers();
+    vi.restoreAllMocks();
+  });
+
+  it('dispatches state on WS message', () => {
+    const { result } = renderHook(() => useGameSocket('game-1'));
+
+    const ws = MockWebSocket.instances[0];
+    const mockState = {
+      game_id: 'game-1', turn: 1, current_phase: 'main1', active_seat: 0,
+      players: [], stack: [], game_over: false, winner: null,
+    };
+
+    act(() => {
+      ws.simulateMessage({ type: 'state', payload: mockState });
+    });
+
+    expect(result.current.gameState).toEqual(mockState);
+  });
+
+  it('dispatches hand cards on WS message', () => {
+    const { result } = renderHook(() => useGameSocket('game-1'));
+    const ws = MockWebSocket.instances[0];
+
+    const cards = [{ name: 'Sol Ring', type: 'Artifact', tapped: false, is_commander: false, oracle: '', cmc: 1 }];
+
+    act(() => {
+      ws.simulateMessage({ type: 'hand', cards });
+    });
+
+    expect(result.current.hand).toEqual(cards);
+  });
+
+  it('sends heartbeat ping every 15s', () => {
+    renderHook(() => useGameSocket('game-1'));
+    const ws = MockWebSocket.instances[0];
+
+    act(() => {
+      vi.advanceTimersByTime(15000);
+    });
+
+    expect(ws.sent).toContain(JSON.stringify({ type: 'ping' }));
+  });
+});

frontend/src/assets/react.svg

@@ -0,0 +1 @@
+<svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" class="iconify iconify--logos" width="35.93" height="32" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 228"><path fill="#00D8FF" d="M210.483 73.824a171.49 171.49 0 0 0-8.24-2.597c.465-1.9.893-3.777 1.273-5.621c6.238-30.281 2.16-54.676-11.769-62.708c-13.355-7.7-35.196.329-57.254 19.526a171.23 171.23 0 0 0-6.375 5.848a155.866 155.866 0 0 0-4.241-3.917C100.759 3.829 77.587-4.822 63.673 3.233C50.33 10.957 46.379 33.89 51.995 62.588a170.974 170.974 0 0 0 1.892 8.48c-3.28.932-6.445 1.924-9.474 2.98C17.309 83.498 0 98.307 0 113.668c0 15.865 18.582 31.778 46.812 41.427a145.52 145.52 0 0 0 6.921 2.165a167.467 167.467 0 0 0-2.01 9.138c-5.354 28.2-1.173 50.591 12.134 58.266c13.744 7.926 36.812-.22 59.273-19.855a145.567 145.567 0 0 0 5.342-4.923a168.064 168.064 0 0 0 6.92 6.314c21.758 18.722 43.246 26.282 56.54 18.586c13.731-7.949 18.194-32.003 12.4-61.268a145.016 145.016 0 0 0-1.535-6.842c1.62-.48 3.21-.974 4.76-1.488c29.348-9.723 48.443-25.443 48.443-41.52c0-15.417-17.868-30.326-45.517-39.844Zm-6.365 70.984c-1.4.463-2.836.91-4.3 1.345c-3.24-10.257-7.612-21.163-12.963-32.432c5.106-11 9.31-21.767 12.459-31.957c2.619.758 5.16 1.557 7.61 2.4c23.69 8.156 38.14 20.213 38.14 29.504c0 9.896-15.606 22.743-40.946 31.14Zm-10.514 20.834c2.562 12.94 2.927 24.64 1.23 33.787c-1.524 8.219-4.59 13.698-8.382 15.893c-8.067 4.67-25.32-1.4-43.927-17.412a156.726 156.726 0 0 1-6.437-5.87c7.214-7.889 14.423-17.06 21.459-27.246c12.376-1.098 24.068-2.894 34.671-5.345a134.17 134.17 0 0 1 1.386 6.193ZM87.276 214.515c-7.882 2.783-14.16 2.863-17.955.675c-8.075-4.657-11.432-22.636-6.853-46.752a156.923 156.923 0 0 1 1.869-8.499c10.486 2.32 22.093 3.988 34.498 4.994c7.084 9.967 14.501 19.128 21.976 27.15a134.668 134.668 0 0 1-4.877 4.492c-9.933 8.682-19.886 14.842-28.658 17.94ZM50.35 144.747c-12.483-4.267-22.792-9.812-29.858-15.863c-6.35-5.437-9.555-10.836-9.555-15.216c0-9.322 13.897-21.212 37.076-29.293c2.813-.98 5.757-1.905 8.812-2.773c3.204 10.42 7.406 21.315 12.477 32.332c-5.137 11.18-9.399 22.249-12.634 32.792a134.718 134.718 0 0 1-6.318-1.979Zm12.378-84.26c-4.811-24.587-1.616-43.134 6.425-47.789c8.564-4.958 27.502 2.111 47.463 19.835a144.318 144.318 0 0 1 3.841 3.545c-7.438 7.987-14.787 17.08-21.808 26.988c-12.04 1.116-23.565 2.908-34.161 5.309a160.342 160.342 0 0 1-1.76-7.887Zm110.427 27.268a347.8 347.8 0 0 0-7.785-12.803c8.168 1.033 15.994 2.404 23.343 4.08c-2.206 7.072-4.956 14.465-8.193 22.045a381.151 381.151 0 0 0-7.365-13.322Zm-45.032-43.861c5.044 5.465 10.096 11.566 15.065 18.186a322.04 322.04 0 0 0-30.257-.006c4.974-6.559 10.069-12.652 15.192-18.18ZM82.802 87.83a323.167 323.167 0 0 0-7.227 13.238c-3.184-7.553-5.909-14.98-8.134-22.152c7.304-1.634 15.093-2.97 23.209-3.984a321.524 321.524 0 0 0-7.848 12.897Zm8.081 65.352c-8.385-.936-16.291-2.203-23.593-3.793c2.26-7.3 5.045-14.885 8.298-22.6a321.187 321.187 0 0 0 7.257 13.246c2.594 4.48 5.28 8.868 8.038 13.147Zm37.542 31.03c-5.184-5.592-10.354-11.779-15.403-18.433c4.902.192 9.899.29 14.978.29c5.218 0 10.376-.117 15.453-.343c-4.985 6.774-10.018 12.97-15.028 18.486Zm52.198-57.817c3.422 7.8 6.306 15.345 8.596 22.52c-7.422 1.694-15.436 3.058-23.88 4.071a382.417 382.417 0 0 0 7.859-13.026a347.403 347.403 0 0 0 7.425-13.565Zm-16.898 8.101a358.557 358.557 0 0 1-12.281 19.815a329.4 329.4 0 0 1-23.444.823c-7.967 0-15.716-.248-23.178-.732a310.202 310.202 0 0 1-12.513-19.846h.001a307.41 307.41 0 0 1-10.923-20.627a310.278 310.278 0 0 1 10.89-20.637l-.001.001a307.318 307.318 0 0 1 12.413-19.761c7.613-.576 15.42-.876 23.31-.876H128c7.926 0 15.743.303 23.354.883a329.357 329.357 0 0 1 12.335 19.695a358.489 358.489 0 0 1 11.036 20.54a329.472 329.472 0 0 1-11 20.722Zm22.56-122.124c8.572 4.944 11.906 24.881 6.52 51.026c-.344 1.668-.73 3.367-1.15 5.09c-10.622-2.452-22.155-4.275-34.23-5.408c-7.034-10.017-14.323-19.124-21.64-27.008a160.789 160.789 0 0 1 5.888-5.4c18.9-16.447 36.564-22.941 44.612-18.3ZM128 90.808c12.625 0 22.86 10.235 22.86 22.86s-10.235 22.86-22.86 22.86s-22.86-10.235-22.86-22.86s10.235-22.86 22.86-22.86Z"></path></svg>

frontend/src/assets/vite.svg

@@ -0,0 +1 @@
+<svg xmlns="http://www.w3.org/2000/svg" width="77" height="47" fill="none" aria-labelledby="vite-logo-title" viewBox="0 0 77 47"><title id="vite-logo-title">Vite</title><style>.parenthesis{fill:#000}@media (prefers-color-scheme:dark){.parenthesis{fill:#fff}}</style><path fill="#9135ff" d="M40.151 45.71c-.663.844-2.02.374-2.02-.699V34.708a2.26 2.26 0 0 0-2.262-2.262H24.493c-.92 0-1.457-1.04-.92-1.788l7.479-10.471c1.07-1.498 0-3.578-1.842-3.578H15.443c-.92 0-1.456-1.04-.92-1.788l9.696-13.576c.213-.297.556-.474.92-.474h28.894c.92 0 1.456 1.04.92 1.788l-7.48 10.472c-1.07 1.497 0 3.578 1.842 3.578h11.376c.944 0 1.474 1.087.89 1.83L40.153 45.712z"/><mask id="a" width="48" height="47" x="14" y="0" maskUnits="userSpaceOnUse" style="mask-type:alpha"><path fill="#000" d="M40.047 45.71c-.663.843-2.02.374-2.02-.699V34.708a2.26 2.26 0 0 0-2.262-2.262H24.389c-.92 0-1.457-1.04-.92-1.788l7.479-10.472c1.07-1.497 0-3.578-1.842-3.578H15.34c-.92 0-1.456-1.04-.92-1.788l9.696-13.575c.213-.297.556-.474.92-.474H53.93c.92 0 1.456 1.04.92 1.788L47.37 13.03c-1.07 1.498 0 3.578 1.842 3.578h11.376c.944 0 1.474 1.088.89 1.831L40.049 45.712z"/></mask><g mask="url(#a)"><g filter="url(#b)"><ellipse cx="5.508" cy="14.704" fill="#eee6ff" rx="5.508" ry="14.704" transform="rotate(269.814 20.96 11.29)scale(-1 1)"/></g><g filter="url(#c)"><ellipse cx="10.399" cy="29.851" fill="#eee6ff" rx="10.399" ry="29.851" transform="rotate(89.814 -16.902 -8.275)scale(1 -1)"/></g><g filter="url(#d)"><ellipse cx="5.508" cy="30.487" fill="#8900ff" rx="5.508" ry="30.487" transform="rotate(89.814 -19.197 -7.127)scale(1 -1)"/></g><g filter="url(#e)"><ellipse cx="5.508" cy="30.599" fill="#8900ff" rx="5.508" ry="30.599" transform="rotate(89.814 -25.928 4.177)scale(1 -1)"/></g><g filter="url(#f)"><ellipse cx="5.508" cy="30.599" fill="#8900ff" rx="5.508" ry="30.599" transform="rotate(89.814 -25.738 5.52)scale(1 -1)"/></g><g filter="url(#g)"><ellipse cx="14.072" cy="22.078" fill="#eee6ff" rx="14.072" ry="22.078" transform="rotate(93.35 31.245 55.578)scale(-1 1)"/></g><g filter="url(#h)"><ellipse cx="3.47" cy="21.501" fill="#8900ff" rx="3.47" ry="21.501" transform="rotate(89.009 35.419 55.202)scale(-1 1)"/></g><g filter="url(#i)"><ellipse cx="3.47" cy="21.501" fill="#8900ff" rx="3.47" ry="21.501" transform="rotate(89.009 35.419 55.202)scale(-1 1)"/></g><g filter="url(#j)"><ellipse cx="14.592" cy="9.743" fill="#8900ff" rx="4.407" ry="29.108" transform="rotate(39.51 14.592 9.743)"/></g><g filter="url(#k)"><ellipse cx="61.728" cy="-5.321" fill="#8900ff" rx="4.407" ry="29.108" transform="rotate(37.892 61.728 -5.32)"/></g><g filter="url(#l)"><ellipse cx="55.618" cy="7.104" fill="#00c2ff" rx="5.971" ry="9.665" transform="rotate(37.892 55.618 7.104)"/></g><g filter="url(#m)"><ellipse cx="12.326" cy="39.103" fill="#8900ff" rx="4.407" ry="29.108" transform="rotate(37.892 12.326 39.103)"/></g><g filter="url(#n)"><ellipse cx="12.326" cy="39.103" fill="#8900ff" rx="4.407" ry="29.108" transform="rotate(37.892 12.326 39.103)"/></g><g filter="url(#o)"><ellipse cx="49.857" cy="30.678" fill="#8900ff" rx="4.407" ry="29.108" transform="rotate(37.892 49.857 30.678)"/></g><g filter="url(#p)"><ellipse cx="52.623" cy="33.171" fill="#00c2ff" rx="5.971" ry="15.297" transform="rotate(37.892 52.623 33.17)"/></g></g><path d="M6.919 0c-9.198 13.166-9.252 33.575 0 46.789h6.215c-9.25-13.214-9.196-33.623 0-46.789zm62.424 0h-6.215c9.198 13.166 9.252 33.575 0 46.789h6.215c9.25-13.214 9.196-33.623 0-46.789" class="parenthesis"/><defs><filter id="b" width="60.045" height="41.654" x="-5.564" y="16.92" color-interpolation-filters="sRGB" filterUnits="userSpaceOnUse"><feFlood flood-opacity="0" result="BackgroundImageFix"/><feBlend in="SourceGraphic" in2="BackgroundImageFix" result="shape"/><feGaussianBlur result="effect1_foregroundBlur_2002_17286" stdDeviation="7.659"/></filter><filter id="c" width="90.34" height="51.437" x="-40.407" y="-6.762" color-interpolation-filters="sRGB" filterUnits="userSpaceOnUse"><feFlood flood-opacity="0" result="BackgroundImageFix"/><feBlend in="SourceGraphic" in2="BackgroundImageFix" result="shape"/><feGaussianBlur result="effect1_foregroundBlur_2002_17286" stdDeviation="7.659"/></filter><filter id="d" width="79.355" height="29.4" x="-35.435" y="2.801" color-interpolation-filters="sRGB" filterUnits="userSpaceOnUse"><feFlood flood-opacity="0" result="BackgroundImageFix"/><feBlend in="SourceGraphic" in2="BackgroundImageFix" result="shape"/><feGaussianBlur result="effect1_foregroundBlur_2002_17286" stdDeviation="4.596"/></filter><filter id="e" width="79.579" height="29.4" x="-30.84" y="20.8" color-interpolation-filters="sRGB" filterUnits="userSpaceOnUse"><feFlood flood-opacity="0" result="BackgroundImageFix"/><feBlend in="SourceGraphic" in2="BackgroundImageFix" result="shape"/><feGaussianBlur result="effect1_foregroundBlur_2002_17286" stdDeviation="4.596"/></filter><filter id="f" width="79.579" height="29.4" x="-29.307" y="21.949" color-interpolation-filters="sRGB" filterUnits="usLine truncated

frontend/src/components/Board.tsx

@@ -0,0 +1,54 @@
+import type { GameState } from '../types/game';
+import { SEAT_COLOURS } from '../constants';
+import PlayerZone from './PlayerZone';
+import CardChip from './CardChip';
+
+interface BoardProps {
+  gameState: GameState;
+  thinkingSeats: Set<number>;
+}
+
+export default function Board({ gameState, thinkingSeats }: BoardProps) {
+  const { players, active_seat, stack } = gameState;
+
+  const getPlayer = (seat: number) => players.find((p) => p.seat === seat);
+
+  const renderZone = (seat: number, className: string) => {
+    const player = getPlayer(seat);
+    if (!player) return <div className={className} />;
+    return (
+      <div className={className}>
+        <PlayerZone
+          player={player}
+          isActive={active_seat === seat}
+          isThinking={thinkingSeats.has(seat)}
+          seatColour={SEAT_COLOURS[seat]}
+        />
+      </div>
+    );
+  };
+
+  return (
+    <div className="grid grid-cols-[180px_1fr_180px] grid-rows-3 gap-1 flex-1">
+      {/* Row 1: P1 (top-left), stack center, P2 (top-right) */}
+      {renderZone(1, 'row-start-1 col-start-1')}
+      <div className="bg-bg-zone border border-accent-purple rounded-md min-h-[60px] row-span-3 col-start-2 p-2 overflow-y-auto">
+        <span className="text-[10px] text-[#666] block mb-1">Stack</span>
+        <div className="flex flex-wrap gap-0.5">
+          {stack.map((card, i) => (
+            <CardChip key={i} card={card} />
+          ))}
+        </div>
+      </div>
+      {renderZone(2, 'row-start-1 col-start-3')}
+
+      {/* Row 2: empty side cells (stack spans all 3 rows in center) */}
+      <div className="row-start-2 col-start-1" />
+      <div className="row-start-2 col-start-3" />
+
+      {/* Row 3: P0 (bottom-left), P3 (bottom-right) */}
+      {renderZone(0, 'row-start-3 col-start-1')}
+      {renderZone(3, 'row-start-3 col-start-3')}
+    </div>
+  );
+}

frontend/src/components/CardChip.tsx

@@ -0,0 +1,20 @@
+import type { Card } from '../types/game';
+
+interface CardChipProps {
+  card: Card;
+  onClick?: () => void;
+}
+
+export default function CardChip({ card, onClick }: CardChipProps) {
+  return (
+    <span
+      onClick={onClick}
+      className={`bg-card-bg border border-[#444] rounded text-[10px] max-w-[110px] overflow-hidden text-ellipsis whitespace-nowrap cursor-pointer px-1.5 py-0.5 hover:border-[#ffdc32] ${
+        card.tapped ? 'opacity-55 italic' : ''
+      }`}
+    >
+      {card.name}
+      {card.is_commander && <span className="text-[#ffd200]"> ★</span>}
+    </span>
+  );
+}

frontend/src/components/GameOverOverlay.tsx

@@ -0,0 +1,40 @@
+import type { Player } from '../types/game';
+import { SEAT_COLOURS } from '../constants';
+
+interface GameOverOverlayProps {
+  show: boolean;
+  winner: Player | null;
+  reason: string;
+  players: Player[];
+}
+
+export default function GameOverOverlay({ show, winner, reason, players }: GameOverOverlayProps) {
+  if (!show) return null;
+
+  return (
+    <div className="fixed inset-0 bg-black/80 z-[100] flex items-center justify-center">
+      <div className="bg-bg-zone border-2 border-[#4060c0] rounded-[10px] px-[60px] py-10 text-center">
+        {winner && (
+          <h2
+            className="text-4xl font-bold"
+            style={{ color: SEAT_COLOURS[winner.seat] }}
+          >
+            {winner.name} Wins!
+          </h2>
+        )}
+        <p className="text-base text-[#aaa] mt-2">{reason}</p>
+        <ul className="mt-6 space-y-1">
+          {players.map((p) => (
+            <li key={p.seat} className="text-sm">
+              <span style={{ color: SEAT_COLOURS[p.seat] }}>P{p.seat}</span>{' '}
+              {p.name} — {p.life} life
+              {p.eliminated && (
+                <span className="text-red-500 ml-1 font-bold">[ELIM]</span>
+              )}
+            </li>
+          ))}
+        </ul>
+      </div>
+    </div>
+  );
+}

frontend/src/components/HandCard.tsx

@@ -0,0 +1,23 @@
+import type { Card } from '../types/game';
+
+interface HandCardProps {
+  card: Card;
+  index: number;
+  isSelected: boolean;
+  onClick: () => void;
+}
+
+export default function HandCard({ card, isSelected, onClick }: HandCardProps) {
+  return (
+    <button
+      onClick={onClick}
+      className={`rounded text-[11px] px-2 py-1 cursor-pointer ${
+        isSelected
+          ? 'border border-[#ffdc32] bg-card-selected'
+          : 'bg-card-hand border border-[#5060a0]'
+      }`}
+    >
+      {card.name}
+    </button>
+  );
+}

frontend/src/components/MoveBar.tsx

@@ -0,0 +1,54 @@
+import { useState } from 'react';
+import type { Card } from '../types/game';
+import HandCard from './HandCard';
+
+interface MoveBarProps {
+  hand: Card[];
+  humanSeat: number;
+  onPass: () => void;
+  onConcede: () => void;
+  onPlayCard: (index: number) => void;
+  status: string;
+}
+
+export default function MoveBar({ hand, onPass, onConcede, onPlayCard, status }: MoveBarProps) {
+  const [selectedIndex, setSelectedIndex] = useState<number | null>(null);
+
+  const handleCardClick = (index: number) => {
+    if (selectedIndex === index) {
+      onPlayCard(index);
+      setSelectedIndex(null);
+    } else {
+      setSelectedIndex(index);
+    }
+  };
+
+  return (
+    <div className="bg-bg-bar border-t border-[#2a2e48] px-2.5 py-1.5 flex items-center gap-2">
+      <button
+        onClick={onPass}
+        className="bg-[#2a5090] border border-[#4080d0] text-white rounded px-3 py-1 text-sm cursor-pointer hover:brightness-110"
+      >
+        Pass Priority
+      </button>
+      <button
+        onClick={() => {
+          if (window.confirm('Are you sure you want to concede?')) onConcede();
+        }}
+        className="bg-[#2a3868] border border-[#4060a0] text-white rounded px-3 py-1 text-sm cursor-pointer"
+      >
+        Concede
+      </button>
+      {hand.map((card, i) => (
+        <HandCard
+          key={i}
+          card={card}
+          index={i}
+          isSelected={selectedIndex === i}
+          onClick={() => handleCardClick(i)}
+        />
+      ))}
+      <span className="text-[11px] text-[#778] ml-auto">{status}</span>
+    </div>
+  );
+}

frontend/src/components/NarrationPanel.tsx

@@ -0,0 +1,68 @@
+import { useRef, useEffect, useState } from 'react';
+import type { NarrationLine } from '../types/game';
+import { SEAT_COLOURS } from '../constants';
+
+interface NarrationPanelProps {
+  lines: NarrationLine[];
+  thinkingSeats: Set<number>;
+}
+
+export default function NarrationPanel({ lines, thinkingSeats }: NarrationPanelProps) {
+  const containerRef = useRef<HTMLDivElement>(null);
+  const [dots, setDots] = useState('.');
+
+  // Auto-scroll on new lines
+  useEffect(() => {
+    const el = containerRef.current;
+    if (el) el.scrollTop = el.scrollHeight;
+  }, [lines.length]);
+
+  // Cycling dots for thinking animation
+  useEffect(() => {
+    if (thinkingSeats.size === 0) return;
+    const interval = setInterval(() => {
+      setDots((prev) => (prev.length >= 3 ? '.' : prev + '.'));
+    }, 500);
+    return () => clearInterval(interval);
+  }, [thinkingSeats.size]);
+
+  const displayed = lines.slice(-120);
+
+  return (
+    <div
+      ref={containerRef}
+      className="w-[260px] bg-bg-panel border-l border-[#2a2e48] overflow-y-auto p-2 flex flex-col"
+    >
+      {displayed.map((line, i) => {
+        let colorClass = 'text-[#e8e8e8]';
+        let style: React.CSSProperties | undefined;
+
+        if (line.isSystem) {
+          colorClass = 'text-[#556]';
+        } else if (line.seat !== undefined) {
+          colorClass = '';
+          style = { color: SEAT_COLOURS[line.seat] };
+        }
+
+        return (
+          <div
+            key={i}
+            className={`text-[11px] leading-[1.4] break-words ${colorClass}`}
+            style={style}
+          >
+            {line.text}
+          </div>
+        );
+      })}
+      {thinkingSeats.size > 0 &&
+        Array.from(thinkingSeats).map((seat) => (
+          <div
+            key={`thinking-${seat}`}
+            className="text-[11px] leading-[1.4] break-words text-accent-think italic"
+          >
+            P{seat} thinking{dots}
+          </div>
+        ))}
+    </div>
+  );
+}

frontend/src/components/PhaseBar.tsx

@@ -0,0 +1,37 @@
+import type { Phase, Player } from '../types/game';
+import { PHASES, SEAT_COLOURS } from '../constants';
+
+interface PhaseBarProps {
+  phase: Phase;
+  activeSeat: number;
+  players: Player[];
+}
+
+export default function PhaseBar({ phase, activeSeat, players }: PhaseBarProps) {
+  const activePlayer = players.find((p) => p.seat === activeSeat);
+
+  return (
+    <div className="h-[34px] bg-[#1e2338] px-2 py-1 flex items-center gap-1">
+      {PHASES.map((p) => (
+        <button
+          key={p}
+          className={`text-[11px] rounded px-2 py-0.5 ${
+            p === phase
+              ? 'bg-accent-blue text-white'
+              : 'bg-[#252a42] text-[#888]'
+          }`}
+        >
+          {p}
+        </button>
+      ))}
+      {activePlayer && (
+        <span
+          className="text-[11px] ml-auto font-bold"
+          style={{ color: SEAT_COLOURS[activeSeat] }}
+        >
+          P{activeSeat}: {activePlayer.name}
+        </span>
+      )}
+    </div>
+  );
+}

frontend/src/components/PlayerZone.tsx

@@ -0,0 +1,40 @@
+import type { Player } from '../types/game';
+import CardChip from './CardChip';
+
+interface PlayerZoneProps {
+  player: Player;
+  isActive: boolean;
+  isThinking: boolean;
+  seatColour: string;
+}
+
+function lifeColor(life: number): string {
+  if (life >= 30) return 'text-life-high';
+  if (life >= 15) return 'text-life-med';
+  return 'text-life-low';
+}
+
+export default function PlayerZone({ player, isActive, isThinking, seatColour }: PlayerZoneProps) {
+  return (
+    <div
+      className={`bg-bg-zone border border-[#3a4060] rounded-md p-1.5 ${
+        isThinking ? 'animate-pulse' : ''
+      } ${player.eliminated ? 'opacity-40' : ''}`}
+      style={isActive ? { borderColor: seatColour } : undefined}
+    >
+      <div className="flex items-center justify-between mb-1">
+        <span className="text-[11px] text-[#666]">
+          P{player.seat}: {player.name}
+        </span>
+        <span className={`text-[28px] font-bold ${lifeColor(player.life)}`}>
+          {player.life}
+        </span>
+      </div>
+      <div className="flex flex-wrap gap-0.5">
+        {player.battlefield.map((card, i) => (
+          <CardChip key={i} card={card} />
+        ))}
+      </div>
+    </div>
+  );
+}

frontend/src/constants.ts

@@ -0,0 +1,9 @@
+import type { Phase } from './types/game';
+
+export const SEAT_COLOURS = ['#4696ff', '#ff6450', '#50d278', '#dcaa32'] as const;
+
+export const PHASES: Phase[] = [
+  'untap', 'upkeep', 'draw', 'main1',
+  'begin_combat', 'declare_attackers', 'declare_blockers', 'combat_damage',
+  'main2', 'end_step', 'cleanup',
+];

frontend/src/hooks/useGameSocket.ts

@@ -0,0 +1,132 @@
+import { useReducer, useRef, useEffect, useCallback } from 'react';
+import { gameReducer, initialState } from '../store/gameReducer';
+import type { WsIncoming, NarrationLine } from '../types/game';
+
+export function useGameSocket(gameId: string) {
+  const [state, dispatch] = useReducer(gameReducer, initialState);
+  const wsRef = useRef<WebSocket | null>(null);
+  const reconnectTimeout = useRef<ReturnType<typeof setTimeout> | null>(null);
+  const heartbeatInterval = useRef<ReturnType<typeof setInterval> | null>(null);
+  const intentionalClose = useRef(false);
+
+  const connect = useCallback(() => {
+    const url = `ws://${window.location.host}/ws/game/${gameId}`;
+    const ws = new WebSocket(url);
+    wsRef.current = ws;
+
+    ws.onopen = () => {
+      dispatch({ type: 'SET_STATUS', status: 'Connected' });
+    };
+
+    ws.onmessage = (event) => {
+      const msg: WsIncoming = JSON.parse(event.data);
+
+      switch (msg.type) {
+        case 'state':
+          dispatch({ type: 'SET_STATE', payload: msg.payload });
+          break;
+
+        case 'hand':
+          dispatch({ type: 'SET_HAND', cards: msg.cards });
+          break;
+
+        case 'event': {
+          const line: NarrationLine = {
+            text: msg.text,
+            seat: msg.seat,
+            isSystem: false,
+            timestamp: Date.now(),
+          };
+          dispatch({ type: 'ADD_NARRATION', line });
+          break;
+        }
+
+        case 'phase': {
+          const line: NarrationLine = {
+            text: `Phase: ${msg.phase} (P${msg.active_seat})`,
+            isSystem: true,
+            timestamp: Date.now(),
+          };
+          dispatch({ type: 'ADD_NARRATION', line });
+          break;
+        }
+
+        case 'thinking':
+          dispatch({ type: 'SET_THINKING', seat: msg.seat, active: msg.active });
+          break;
+
+        case 'game_over':
+          dispatch({ type: 'GAME_OVER', winner: msg.winner, reason: msg.reason });
+          break;
+
+        case 'pong':
+          break;
+
+        case 'error':
+          dispatch({ type: 'SET_STATUS', status: `Error: ${msg.message}` });
+          break;
+      }
+    };
+
+    ws.onclose = () => {
+      if (!intentionalClose.current) {
+        dispatch({ type: 'SET_STATUS', status: 'Disconnected — reconnecting…' });
+        reconnectTimeout.current = setTimeout(() => {
+          connect();
+        }, 3000);
+      }
+    };
+
+    ws.onerror = () => {
+      dispatch({ type: 'SET_STATUS', status: 'Connection error' });
+    };
+  }, [gameId]);
+
+  // Connect on mount
+  useEffect(() => {
+    intentionalClose.current = false;
+    connect();
+
+    return () => {
+      intentionalClose.current = true;
+      if (reconnectTimeout.current) clearTimeout(reconnectTimeout.current);
+      wsRef.current?.close();
+    };
+  }, [connect]);
+
+  // Heartbeat ping every 15s
+  useEffect(() => {
+    heartbeatInterval.current = setInterval(() => {
+      if (wsRef.current?.readyState === WebSocket.OPEN) {
+        wsRef.current.send(JSON.stringify({ type: 'ping' }));
+      }
+    }, 15000);
+
+    return () => {
+      if (heartbeatInterval.current) clearInterval(heartbeatInterval.current);
+    };
+  }, []);
+
+  const sendMove = useCallback((move: string, data?: Record<string, unknown>) => {
+    if (wsRef.current?.readyState === WebSocket.OPEN) {
+      wsRef.current.send(JSON.stringify({ type: 'move', move, ...data }));
+    }
+  }, []);
+
+  const sendConcede = useCallback(() => {
+    if (wsRef.current?.readyState === WebSocket.OPEN) {
+      wsRef.current.send(JSON.stringify({ type: 'concede' }));
+    }
+  }, []);
+
+  return {
+    gameState: state.gameState,
+    narration: state.narration,
+    thinking: state.thinkingSeats,
+    hand: state.hand,
+    gameOver: state.gameOver,
+    status: state.status,
+    sendMove,
+    sendConcede,
+  };
+}

frontend/src/hooks/useLobbySock.ts

@@ -0,0 +1,163 @@
+import { useReducer, useRef, useEffect, useCallback } from 'react';
+
+export interface Seat {
+  seat: number;
+  name: string;
+  ready: boolean;
+  isAi: boolean;
+}
+
+export interface RoomState {
+  roomId: string;
+  roomName: string;
+  host: string;
+  seats: (Seat | null)[];
+}
+
+export interface ChatMessage {
+  role: 'PLAYER' | 'AI' | 'SYSTEM';
+  name: string;
+  text: string;
+  timestamp: number;
+}
+
+type LobbyAction =
+  | { type: 'SET_ROOM'; payload: RoomState }
+  | { type: 'ADD_CHAT'; message: ChatMessage }
+  | { type: 'SET_CHAT_HISTORY'; messages: ChatMessage[] }
+  | { type: 'LAUNCH'; gameId: string };
+
+interface LobbyState {
+  roomState: RoomState | null;
+  chat: ChatMessage[];
+  launchedGameId: string | null;
+}
+
+const initialState: LobbyState = {
+  roomState: null,
+  chat: [],
+  launchedGameId: null,
+};
+
+function lobbyReducer(state: LobbyState, action: LobbyAction): LobbyState {
+  switch (action.type) {
+    case 'SET_ROOM':
+      return { ...state, roomState: action.payload };
+    case 'ADD_CHAT':
+      return { ...state, chat: [...state.chat, action.message].slice(-50) };
+    case 'SET_CHAT_HISTORY':
+      return { ...state, chat: action.messages.slice(-50) };
+    case 'LAUNCH':
+      return { ...state, launchedGameId: action.gameId };
+    default:
+      return state;
+  }
+}
+
+export function useLobbySock(roomId: string, playerName: string) {
+  const [state, dispatch] = useReducer(lobbyReducer, initialState);
+  const wsRef = useRef<WebSocket | null>(null);
+  const reconnectTimeout = useRef<ReturnType<typeof setTimeout> | null>(null);
+  const heartbeatInterval = useRef<ReturnType<typeof setInterval> | null>(null);
+  const intentionalClose = useRef(false);
+
+  const connect = useCallback(() => {
+    const url = `ws://${window.location.host}/ws/lobby/${roomId}?name=${encodeURIComponent(playerName)}`;
+    const ws = new WebSocket(url);
+    wsRef.current = ws;
+
+    ws.onmessage = (event) => {
+      const msg = JSON.parse(event.data);
+
+      switch (msg.type) {
+        case 'lobby_state':
+          dispatch({ type: 'SET_ROOM', payload: msg.payload });
+          break;
+        case 'chat':
+          dispatch({
+            type: 'ADD_CHAT',
+            message: {
+              role: msg.role ?? 'PLAYER',
+              name: msg.name ?? '',
+              text: msg.text,
+              timestamp: Date.now(),
+            },
+          });
+          break;
+        case 'chat_history':
+          dispatch({
+            type: 'SET_CHAT_HISTORY',
+            messages: (msg.messages ?? []).map((m: { role?: string; name?: string; text: string }) => ({
+              role: m.role ?? 'SYSTEM',
+              name: m.name ?? '',
+              text: m.text,
+              timestamp: Date.now(),
+            })),
+          });
+          break;
+        case 'launch':
+          dispatch({ type: 'LAUNCH', gameId: msg.game_id });
+          break;
+        case 'pong':
+          break;
+      }
+    };
+
+    ws.onclose = () => {
+      if (!intentionalClose.current) {
+        reconnectTimeout.current = setTimeout(() => connect(), 3000);
+      }
+    };
+  }, [roomId, playerName]);
+
+  useEffect(() => {
+    intentionalClose.current = false;
+    connect();
+
+    return () => {
+      intentionalClose.current = true;
+      if (reconnectTimeout.current) clearTimeout(reconnectTimeout.current);
+      wsRef.current?.close();
+    };
+  }, [connect]);
+
+  // Heartbeat ping every 15s
+  useEffect(() => {
+    heartbeatInterval.current = setInterval(() => {
+      if (wsRef.current?.readyState === WebSocket.OPEN) {
+        wsRef.current.send(JSON.stringify({ type: 'ping' }));
+      }
+    }, 15000);
+
+    return () => {
+      if (heartbeatInterval.current) clearInterval(heartbeatInterval.current);
+    };
+  }, []);
+
+  const sendReady = useCallback((seat: number) => {
+    if (wsRef.current?.readyState === WebSocket.OPEN) {
+      wsRef.current.send(JSON.stringify({ type: 'ready', seat }));
+    }
+  }, []);
+
+  const sendChat = useCallback((text: string) => {
+    if (wsRef.current?.readyState === WebSocket.OPEN) {
+      wsRef.current.send(JSON.stringify({ type: 'chat', text }));
+    }
+  }, []);
+
+  const sendLaunch = useCallback(() => {
+    if (wsRef.current?.readyState === WebSocket.OPEN) {
+      wsRef.current.send(JSON.stringify({ type: 'launch' }));
+    }
+  }, []);
+
+  return {
+    roomState: state.roomState,
+    chat: state.chat,
+    launchedGameId: state.launchedGameId,
+    sendReady,
+    sendChat,
+    sendLaunch,
+  };
+}

frontend/src/index.css

@@ -0,0 +1,3 @@
+@tailwind base;
+@tailwind components;
+@tailwind utilities;

frontend/src/main.tsx

@@ -0,0 +1,10 @@
+import { StrictMode } from 'react'
+import { createRoot } from 'react-dom/client'
+import './index.css'
+import App from './App.tsx'
+
+createRoot(document.getElementById('root')!).render(
+  <StrictMode>
+    <App />
+  </StrictMode>,
+)

frontend/src/pages/GamePage.tsx

@@ -0,0 +1,53 @@
+import { useParams } from 'react-router-dom';
+import { useGameSocket } from '../hooks/useGameSocket';
+import PhaseBar from '../components/PhaseBar';
+import Board from '../components/Board';
+import NarrationPanel from '../components/NarrationPanel';
+import MoveBar from '../components/MoveBar';
+import GameOverOverlay from '../components/GameOverOverlay';
+
+export default function GamePage() {
+  const { gameId } = useParams<{ gameId: string }>();
+  const { gameState, narration, thinking, hand, gameOver, status, sendMove, sendConcede } =
+    useGameSocket(gameId!);
+
+  if (!gameState) {
+    return (
+      <div className="flex items-center justify-center h-screen bg-bg-base text-white">
+        Connecting…
+      </div>
+    );
+  }
+
+  const winnerPlayer = gameOver
+    ? gameState.players.find((p) => p.seat === gameOver.winner) ?? null
+    : null;
+
+  return (
+    <div className="flex flex-col h-screen overflow-hidden bg-bg-base text-white">
+      <PhaseBar
+        phase={gameState.current_phase}
+        activeSeat={gameState.active_seat}
+        players={gameState.players}
+      />
+      <div className="flex flex-1 overflow-hidden">
+        <Board gameState={gameState} thinkingSeats={thinking} />
+        <NarrationPanel lines={narration} thinkingSeats={thinking} />
+      </div>
+      <MoveBar
+        hand={hand}
+        humanSeat={0}
+        onPass={() => sendMove('pass')}
+        onConcede={sendConcede}
+        onPlayCard={(index) => sendMove('play_card', { index })}
+        status={status}
+      />
+      <GameOverOverlay
+        show={gameOver !== null}
+        winner={winnerPlayer}
+        reason={gameOver?.reason ?? ''}
+        players={gameState.players}
+      />
+    </div>
+  );
+}

frontend/src/pages/LobbyPage.tsx

@@ -0,0 +1,141 @@
+import { useEffect, useState } from 'react';
+import { useNavigate } from 'react-router-dom';
+
+interface RoomSummary {
+  roomId: string;
+  name: string;
+  host: string;
+  playerCount: number;
+}
+
+export default function LobbyPage() {
+  const navigate = useNavigate();
+  const [rooms, setRooms] = useState<RoomSummary[]>([]);
+  const [loading, setLoading] = useState(true);
+
+  // Create room form state
+  const [roomName, setRoomName] = useState('');
+  const [hostName, setHostName] = useState('');
+  const [aiSlots, setAiSlots] = useState(0);
+  const [password, setPassword] = useState('');
+  const [creating, setCreating] = useState(false);
+
+  useEffect(() => {
+    fetch('/api/v1/lobby/rooms')
+      .then((res) => res.json())
+      .then((data) => setRooms(data))
+      .catch(() => setRooms([]))
+      .finally(() => setLoading(false));
+  }, []);
+
+  const handleCreate = async (e: React.FormEvent) => {
+    e.preventDefault();
+    setCreating(true);
+    try {
+      const res = await fetch('/api/v1/lobby/rooms', {
+        method: 'POST',
+        headers: { 'Content-Type': 'application/json' },
+        body: JSON.stringify({
+          name: roomName,
+          host: hostName,
+          ai_slots: aiSlots,
+          password: password || undefined,
+        }),
+      });
+      const data = await res.json();
+      navigate(`/room/${data.roomId}?name=${encodeURIComponent(hostName)}`);
+    } catch {
+      setCreating(false);
+    }
+  };
+
+  return (
+    <div className="min-h-screen bg-bg-base text-white p-6">
+      <h1 className="text-3xl font-bold mb-6">Commander AI Lab — Lobby</h1>
+
+      <div className="grid grid-cols-1 md:grid-cols-2 gap-6 max-w-4xl">
+        {/* Room List */}
+        <div className="bg-bg-zone rounded-lg border border-border-zone p-4">
+          <h2 className="text-lg font-semibold mb-3">Open Rooms</h2>
+          {loading ? (
+            <p className="text-text-muted text-sm">Loading...</p>
+          ) : rooms.length === 0 ? (
+            <p className="text-text-muted text-sm">No rooms available. Create one!</p>
+          ) : (
+            <ul className="space-y-2">
+              {rooms.map((room) => (
+                <li key={room.roomId}>
+                  <button
+                    onClick={() => navigate(`/room/${room.roomId}`)}
+                    className="w-full text-left bg-bg-panel border border-border-zone rounded px-3 py-2 hover:border-accent-blue transition-colors"
+                  >
+                    <span className="font-medium">{room.name}</span>
+                    <span className="text-text-muted text-sm ml-2">
+                      Host: {room.host} · {room.playerCount}/4 players
+                    </span>
+                  </button>
+                </li>
+              ))}
+            </ul>
+          )}
+        </div>
+
+        {/* Create Room Form */}
+        <div className="bg-bg-zone rounded-lg border border-border-zone p-4">
+          <h2 className="text-lg font-semibold mb-3">Create Room</h2>
+          <form onSubmit={handleCreate} className="space-y-3">
+            <div>
+              <label className="block text-sm text-text-muted mb-1">Room Name</label>
+              <input
+                type="text"
+                value={roomName}
+                onChange={(e) => setRoomName(e.target.value)}
+                required
+                className="w-full bg-bg-panel border border-border-zone rounded px-3 py-1.5 text-sm text-white focus:outline-none focus:border-accent-blue"
+              />
+            </div>
+            <div>
+              <label className="block text-sm text-text-muted mb-1">Your Name</label>
+              <input
+                type="text"
+                value={hostName}
+                onChange={(e) => setHostName(e.target.value)}
+                required
+                className="w-full bg-bg-panel border border-border-zone rounded px-3 py-1.5 text-sm text-white focus:outline-none focus:border-accent-blue"
+              />
+            </div>
+            <div>
+              <label className="block text-sm text-text-muted mb-1">AI Players</label>
+              <select
+                value={aiSlots}
+                onChange={(e) => setAiSlots(Number(e.target.value))}
+                className="w-full bg-bg-panel border border-border-zone rounded px-3 py-1.5 text-sm text-white focus:outline-none focus:border-accent-blue"
+              >
+                <option value={0}>0 AI players</option>
+                <option value={1}>1 AI player</option>
+                <option value={2}>2 AI players</option>
+                <option value={3}>3 AI players</option>
+              </select>
+            </div>
+            <div>
+              <label className="block text-sm text-text-muted mb-1">Password (optional)</label>
+              <input
+                type="password"
+                value={password}
+                onChange={(e) => setPassword(e.target.value)}
+                className="w-full bg-bg-panel border border-border-zone rounded px-3 py-1.5 text-sm text-white focus:outline-none focus:border-accent-blue"
+              />
+            </div>
+            <button
+              type="submit"
+              disabled={creating}
+              className="w-full bg-btn-primary border border-border-btn-primary text-white rounded px-4 py-2 text-sm font-medium hover:brightness-110 disabled:opacity-50"
+            >
+              {creating ? 'Creating...' : 'Create Room'}
+            </button>
+          </form>
+        </div>
+      </div>
+    </div>
+  );
+}

frontend/src/pages/RoomPage.tsx

@@ -0,0 +1,220 @@
+import { useState, useEffect, useRef } from 'react';
+import { useParams, useSearchParams, useNavigate } from 'react-router-dom';
+import { useLobbySock } from '../hooks/useLobbySock';
+import type { ChatMessage } from '../hooks/useLobbySock';
+
+const ROLE_COLORS: Record<string, string> = {
+  PLAYER: 'text-accent-blue',
+  AI: 'text-accent-purple',
+  SYSTEM: 'text-text-muted',
+};
+
+function ChatBadge({ role }: { role: string }) {
+  return (
+    <span className={`text-[10px] font-bold mr-1 ${ROLE_COLORS[role] ?? 'text-text-muted'}`}>
+      [{role}]
+    </span>
+  );
+}
+
+export default function RoomPage() {
+  const { roomId } = useParams<{ roomId: string }>();
+  const [searchParams] = useSearchParams();
+  const navigate = useNavigate();
+
+  const isSpectator = searchParams.get('spectate') === '1';
+  const nameFromUrl = searchParams.get('name') ?? '';
+  const [playerName, setPlayerName] = useState(nameFromUrl);
+  const [nameConfirmed, setNameConfirmed] = useState(!!nameFromUrl);
+  const [chatInput, setChatInput] = useState('');
+  const chatEndRef = useRef<HTMLDivElement>(null);
+
+  const { roomState, chat, launchedGameId, sendReady, sendChat, sendLaunch } =
+    useLobbySock(roomId!, nameConfirmed ? playerName : '');
+
+  // Redirect when game launches
+  useEffect(() => {
+    if (launchedGameId) {
+      navigate(`/game/${launchedGameId}`);
+    }
+  }, [launchedGameId, navigate]);
+
+  // Auto-scroll chat
+  useEffect(() => {
+    chatEndRef.current?.scrollIntoView({ behavior: 'smooth' });
+  }, [chat.length]);
+
+  // Determine own seat
+  const mySeat = roomState?.seats.findIndex(
+    (s) => s && !s.isAi && s.name === playerName,
+  ) ?? -1;
+  const isHost = mySeat === 0;
+
+  if (!nameConfirmed) {
+    return (
+      <div className="flex items-center justify-center h-screen bg-bg-base text-white">
+        <div className="bg-bg-zone border border-border-zone rounded-lg p-6 w-80">
+          <h2 className="text-lg font-semibold mb-3">Enter your name</h2>
+          <input
+            type="text"
+            value={playerName}
+            onChange={(e) => setPlayerName(e.target.value)}
+            onKeyDown={(e) => {
+              if (e.key === 'Enter' && playerName.trim()) setNameConfirmed(true);
+            }}
+            className="w-full bg-bg-panel border border-border-zone rounded px-3 py-1.5 text-sm text-white mb-3 focus:outline-none focus:border-accent-blue"
+            autoFocus
+          />
+          <button
+            onClick={() => playerName.trim() && setNameConfirmed(true)}
+            className="w-full bg-btn-primary border border-border-btn-primary text-white rounded px-4 py-2 text-sm font-medium hover:brightness-110"
+          >
+            Join Room
+          </button>
+        </div>
+      </div>
+    );
+  }
+
+  if (!roomState) {
+    return (
+      <div className="flex items-center justify-center h-screen bg-bg-base text-white">
+        Connecting to room...
+      </div>
+    );
+  }
+
+  const handleSendChat = () => {
+    if (chatInput.trim()) {
+      sendChat(chatInput.trim());
+      setChatInput('');
+    }
+  };
+
+  return (
+    <div className="flex flex-col h-screen bg-bg-base text-white">
+      {/* Header */}
+      <div className="bg-bg-bar border-b border-border-sep px-4 py-2 flex items-center justify-between">
+        <div>
+          <h1 className="text-lg font-bold">{roomState.roomName}</h1>
+          <span className="text-text-muted text-sm">Host: {roomState.host}</span>
+        </div>
+        {isSpectator && (
+          <span className="text-text-muted text-xs bg-bg-panel px-2 py-1 rounded">
+            SPECTATING
+          </span>
+        )}
+      </div>
+
+      <div className="flex flex-1 overflow-hidden">
+        {/* Seat Grid */}
+        <div className="flex-1 p-4">
+          <div className="grid grid-cols-2 gap-3 max-w-lg">
+            {[0, 1, 2, 3].map((seatIdx) => {
+              const seat = roomState.seats[seatIdx];
+              return (
+                <div
+                  key={seatIdx}
+                  className="bg-bg-zone border border-border-zone rounded-lg p-4 min-h-[100px] flex flex-col items-center justify-center"
+                >
+                  {seat ? (
+                    <>
+                      <span className="text-sm font-medium">
+                        {seat.isAi ? '🤖 AI' : seat.name}
+                      </span>
+                      <span className="text-xs text-text-muted mt-1">
+                        Seat {seatIdx}
+                      </span>
+                      {!seat.isAi && (
+                        <span
+                          className={`text-xs mt-1 font-bold ${
+                            seat.ready ? 'text-life-high' : 'text-life-med'
+                          }`}
+                        >
+                          {seat.ready ? 'READY' : 'Not Ready'}
+                        </span>
+                      )}
+                    </>
+                  ) : (
+                    <>
+                      <span className="text-text-muted text-sm">Waiting...</span>
+                      <span className="text-xs text-text-muted mt-1">
+                        Seat {seatIdx}
+                      </span>
+                    </>
+                  )}
+                </div>
+              );
+            })}
+          </div>
+
+          {/* Action buttons */}
+          {!isSpectator && (
+            <div className="mt-4 flex gap-3">
+              {mySeat >= 0 && (
+                <button
+                  onClick={() => sendReady(mySeat)}
+                  className="bg-btn-primary border border-border-btn-primary text-white rounded px-4 py-2 text-sm font-medium hover:brightness-110"
+                >
+                  Toggle Ready
+                </button>
+              )}
+              {isHost && (
+                <button
+                  onClick={async () => {
+                    try {
+                      const res = await fetch(`/api/v1/lobby/rooms/${roomId}/launch`, {
+                        method: 'POST',
+                      });
+                      const data = await res.json();
+                      if (data.gameId) {
+                        navigate(`/game/${data.gameId}`);
+                      }
+                    } catch {
+                      // launch via WS fallback
+                      sendLaunch();
+                    }
+                  }}
+                  className="bg-life-high border border-life-high text-white rounded px-4 py-2 text-sm font-bold hover:brightness-110"
+                >
+                  Launch Game
+                </button>
+              )}
+            </div>
+          )}
+        </div>
+
+        {/* Chat Panel */}
+        <div className="w-[280px] bg-bg-panel border-l border-border-sep flex flex-col">
+          <div className="px-3 py-2 border-b border-border-sep text-sm font-semibold">
+            Chat
+          </div>
+          <div className="flex-1 overflow-y-auto p-2 space-y-1">
+            {chat.map((msg: ChatMessage, i: number) => (
+              <div key={i} className="text-[11px] leading-[1.4] break-words">
+                <ChatBadge role={msg.role} />
+                {msg.name && <span className="font-medium mr-1">{msg.name}:</span>}
+                <span className="text-text-body">{msg.text}</span>
+              </div>
+            ))}
+            <div ref={chatEndRef} />
+          </div>
+          {!isSpectator && (
+            <div className="border-t border-border-sep p-2">
+              <input
+                type="text"
+                value={chatInput}
+                onChange={(e) => setChatInput(e.target.value)}
+                onKeyDown={(e) => {
+                  if (e.key === 'Enter') handleSendChat();
+                }}
+                placeholder="Type a message..."
+                className="w-full bg-bg-base border border-border-zone rounded px-2 py-1 text-xs text-white focus:outline-none focus:border-accent-blue"
+              />
+            </div>
+          )}
+        </div>
+      </div>
+    </div>
+  );
+}

frontend/src/pages/SimulatorPage.tsx

@@ -0,0 +1,391 @@
+import { useEffect, useRef, useState } from 'react';
+import { SEAT_COLOURS } from '../constants';
+import type { SimDeck, SimResult } from '../types/simulator';
+
+const SEAT_LABELS = ['Seat 1 (You)', 'Seat 2', 'Seat 3', 'Seat 4'];
+
+type SimStatus = 'idle' | 'running' | 'complete' | 'error';
+
+export default function SimulatorPage() {
+  // Deck list from API
+  const [decks, setDecks] = useState<SimDeck[]>([]);
+  const [loadingDecks, setLoadingDecks] = useState(true);
+
+  // Slot selections (4 slots, null = empty)
+  const [slots, setSlots] = useState<(number | null)[]>([null, null, null, null]);
+
+  // Config
+  const [numGames, setNumGames] = useState(10);
+  const [engineType, setEngineType] = useState('heuristic');
+  const [recordLogs, setRecordLogs] = useState(true);
+
+  // Sim state
+  const [simId, setSimId] = useState<string | null>(null);
+  const [status, setStatus] = useState<SimStatus>('idle');
+  const [completed, setCompleted] = useState(0);
+  const [total, setTotal] = useState(0);
+  const [result, setResult] = useState<SimResult | null>(null);
+  const [games, setGames] = useState<unknown[]>([]);
+  const [error, setError] = useState<string | null>(null);
+  const [showLogs, setShowLogs] = useState(false);
+
+  const pollRef = useRef<ReturnType<typeof setInterval> | null>(null);
+
+  // Fetch decks on mount
+  useEffect(() => {
+    fetch('/api/decks')
+      .then((res) => res.json())
+      .then((data) => {
+        const list = (data.decks || []).map((d: Record<string, unknown>) => ({
+          id: d.id as number,
+          name: d.name as string,
+          commander_name: (d.commander_name || '') as string,
+          color_identity: (d.color_identity || []) as string[],
+        }));
+        setDecks(list);
+      })
+      .catch(() => setDecks([]))
+      .finally(() => setLoadingDecks(false));
+  }, []);
+
+  // Polling
+  useEffect(() => {
+    if (status !== 'running' || !simId) return;
+    pollRef.current = setInterval(async () => {
+      try {
+        const res = await fetch(`/api/sim/status?simId=${simId}`);
+        const data = await res.json();
+        setCompleted(data.completed ?? 0);
+        setTotal(data.total ?? 0);
+        if (data.status === 'complete') {
+          setStatus('complete');
+          // Fetch results
+          const rres = await fetch(`/api/sim/result?simId=${simId}`);
+          const rdata = await rres.json();
+          setResult(rdata.summary ?? null);
+          setGames(rdata.games ?? []);
+        } else if (data.status === 'error') {
+          setStatus('error');
+          setError(data.error || 'Simulation failed');
+        }
+      } catch {
+        // keep polling
+      }
+    }, 1000);
+    return () => {
+      if (pollRef.current) clearInterval(pollRef.current);
+    };
+  }, [status, simId]);
+
+  const selectedCount = slots.filter((s) => s !== null).length;
+  const canRun = selectedCount >= 2 && status !== 'running';
+
+  const handleRun = async () => {
+    const deckIds = slots.filter((s): s is number => s !== null);
+    setStatus('running');
+    setCompleted(0);
+    setTotal(numGames);
+    setResult(null);
+    setGames([]);
+    setError(null);
+
+    try {
+      const res = await fetch('/api/sim/run-n-player', {
+        method: 'POST',
+        headers: { 'Content-Type': 'application/json' },
+        body: JSON.stringify({
+          deckIds,
+          numGames,
+          recordLogs,
+          engineType,
+        }),
+      });
+      const data = await res.json();
+      if (!res.ok) {
+        setStatus('error');
+        setError(data.error || 'Failed to start simulation');
+        return;
+      }
+      setSimId(data.simId);
+      setTotal(data.total);
+    } catch {
+      setStatus('error');
+      setError('Network error');
+    }
+  };
+
+  const setSlot = (idx: number, value: number | null) => {
+    setSlots((prev) => {
+      const next = [...prev];
+      next[idx] = value;
+      return next;
+    });
+  };
+
+  const getDeckById = (id: number) => decks.find((d) => d.id === id);
+
+  return (
+    <div className="min-h-screen bg-bg-base text-white p-6">
+      <div className="max-w-5xl mx-auto">
+        <h1 className="text-3xl font-bold mb-2">Batch Simulator</h1>
+        <p className="text-text-muted text-sm mb-6">
+          Run N-player Monte Carlo simulations with your decks.
+        </p>
+
+        {/* Deck Selection */}
+        <div className="mb-6">
+          <h2 className="text-lg font-semibold mb-3">Deck Slots</h2>
+          <div className="grid grid-cols-2 md:grid-cols-4 gap-4">
+            {SEAT_LABELS.map((label, idx) => {
+              const selected = slots[idx];
+              const deck = selected !== null ? getDeckById(selected) : null;
+              return (
+                <div
+                  key={idx}
+                  className="bg-bg-zone rounded-md p-4"
+                  style={{
+                    border: `2px solid ${selected !== null ? SEAT_COLOURS[idx] : '#3a4060'}`,
+                    boxShadow: selected !== null ? `0 0 8px ${SEAT_COLOURS[idx]}40` : 'none',
+                  }}
+                >
+                  <div
+                    className="text-sm font-medium mb-2"
+                    style={{ color: SEAT_COLOURS[idx] }}
+                  >
+                    {label}
+                  </div>
+                  <select
+                    value={selected ?? ''}
+                    onChange={(e) => setSlot(idx, e.target.value ? Number(e.target.value) : null)}
+                    className="w-full bg-bg-panel border border-border-zone rounded px-2 py-1.5 text-sm text-white focus:outline-none focus:border-accent-blue"
+                  >
+                    <option value="">— Empty —</option>
+                    {decks.map((d) => (
+                      <option key={d.id} value={d.id}>
+                        {d.name}
+                      </option>
+                    ))}
+                  </select>
+                  {deck && (
+                    <div className="mt-2 text-xs text-text-muted">
+                      {deck.commander_name && (
+                        <div className="truncate">{deck.commander_name}</div>
+                      )}
+                      {deck.color_identity.length > 0 && (
+                        <div>{deck.color_identity.join(' ')}</div>
+                      )}
+                    </div>
+                  )}
+                </div>
+              );
+            })}
+          </div>
+          {loadingDecks && (
+            <p className="text-text-muted text-sm mt-2">Loading decks...</p>
+          )}
+        </div>
+
+        {/* Config Row */}
+        <div className="flex flex-wrap items-end gap-4 mb-6 bg-bg-zone rounded-md border border-border-zone p-4">
+          <div>
+            <label className="block text-sm text-text-muted mb-1">Games</label>
+            <input
+              type="number"
+              min={1}
+              max={1000}
+              value={numGames}
+              onChange={(e) => setNumGames(Math.max(1, Math.min(1000, Number(e.target.value))))}
+              className="w-24 bg-bg-panel border border-border-zone rounded px-2 py-1.5 text-sm text-white focus:outline-none focus:border-accent-blue"
+            />
+          </div>
+          <div>
+            <label className="block text-sm text-text-muted mb-1">Engine</label>
+            <select
+              value={engineType}
+              onChange={(e) => setEngineType(e.target.value)}
+              className="bg-bg-panel border border-border-zone rounded px-2 py-1.5 text-sm text-white focus:outline-none focus:border-accent-blue"
+            >
+              <option value="heuristic">Heuristic AI</option>
+              <option value="deepseek">DeepSeek AI</option>
+            </select>
+          </div>
+          <div className="flex items-center gap-2">
+            <input
+              type="checkbox"
+              id="recordLogs"
+              checked={recordLogs}
+              onChange={(e) => setRecordLogs(e.target.checked)}
+              className="rounded"
+            />
+            <label htmlFor="recordLogs" className="text-sm text-text-muted">
+              Record Logs
+            </label>
+          </div>
+          <button
+            onClick={handleRun}
+            disabled={!canRun}
+            className="bg-accent-blue text-white rounded px-4 py-2 text-sm font-medium hover:brightness-110 disabled:opacity-50 disabled:cursor-not-allowed"
+          >
+            {status === 'running' ? (
+              <span className="flex items-center gap-2">
+                <span className="inline-block w-4 h-4 border-2 border-white border-t-transparent rounded-full animate-spin" />
+                Running...
+              </span>
+            ) : (
+              'Run Simulation'
+            )}
+          </button>
+        </div>
+
+        {/* Progress */}
+        {status === 'running' && (
+          <div className="mb-6">
+            <div className="flex justify-between text-sm text-text-muted mb-1">
+              <span>Progress</span>
+              <span>
+                {completed} / {total} games
+              </span>
+            </div>
+            <div className="w-full bg-bg-panel rounded-full h-3 overflow-hidden">
+              <div
+                className="h-full bg-accent-blue rounded-full transition-all duration-300"
+                style={{ width: `${total > 0 ? (completed / total) * 100 : 0}%` }}
+              />
+            </div>
+          </div>
+        )}
+
+        {/* Error */}
+        {status === 'error' && error && (
+          <div className="mb-6 bg-red-900/30 border border-red-700 rounded-md p-3 text-sm text-red-300">
+            {error}
+          </div>
+        )}
+
+        {/* Results */}
+        {status === 'complete' && result && (
+          <div className="space-y-6">
+            <div className="flex items-center gap-4 text-sm text-text-muted">
+              <span>{result.totalGames} games</span>
+              <span>{result.playerCount} players</span>
+              <span>{result.elapsedSeconds}s elapsed</span>
+            </div>
+
+            {/* Win Rate Bar Chart */}
+            <div className="bg-bg-zone rounded-md border border-border-zone p-4">
+              <h3 className="text-sm font-semibold mb-3">Win Rates</h3>
+              <div className="space-y-2">
+                {result.players.map((p) => (
+                  <div key={p.seat} className="flex items-center gap-3">
+                    <div className="w-28 text-sm truncate" style={{ color: SEAT_COLOURS[p.seat] }}>
+                      {p.deckName}
+                    </div>
+                    <div className="flex-1 bg-bg-panel rounded-full h-5 overflow-hidden">
+                      <div
+                        className="h-full rounded-full transition-all duration-500"
+                        style={{
+                          width: `${p.winRate}%`,
+                          backgroundColor: SEAT_COLOURS[p.seat],
+                          minWidth: p.winRate > 0 ? '2px' : '0',
+                        }}
+                      />
+                    </div>
+                    <div className="w-16 text-right text-sm font-mono">
+                      {p.winRate}%
+                    </div>
+                  </div>
+                ))}
+              </div>
+            </div>
+
+            {/* Stats Table */}
+            <div className="bg-bg-zone rounded-md border border-border-zone overflow-x-auto">
+              <table className="w-full text-sm">
+                <thead>
+                  <tr className="border-b border-border-zone text-text-muted">
+                    <th className="text-left px-3 py-2">Seat</th>
+                    <th className="text-left px-3 py-2">Deck</th>
+                    <th className="text-right px-3 py-2">Wins</th>
+                    <th className="text-right px-3 py-2">Win%</th>
+                    <th className="text-right px-3 py-2">Avg Turns</th>
+                    <th className="text-right px-3 py-2">Avg Dmg</th>
+                    <th className="text-right px-3 py-2">Avg Spells</th>
+                    <th className="text-right px-3 py-2">Avg Creatures</th>
+                    <th className="text-right px-3 py-2">Avg Removal</th>
+                  </tr>
+                </thead>
+                <tbody>
+                  {result.players.map((p) => {
+                    const isWinner = result.players.every((o) => o.wins <= p.wins);
+                    return (
+                      <tr
+                        key={p.seat}
+                        className={`border-b border-border-zone ${isWinner ? 'font-semibold' : ''}`}
+                        style={{
+                          backgroundColor: `${SEAT_COLOURS[p.seat]}${isWinner ? '20' : '0a'}`,
+                        }}
+                      >
+                        <td className="px-3 py-2" style={{ color: SEAT_COLOURS[p.seat] }}>
+                          P{p.seat + 1}
+                        </td>
+                        <td className="px-3 py-2">{p.deckName}</td>
+                        <td className="text-right px-3 py-2">{p.wins}</td>
+                        <td className="text-right px-3 py-2">{p.winRate}%</td>
+                        <td className="text-right px-3 py-2">{p.avgTurns}</td>
+                        <td className="text-right px-3 py-2">{p.avgDamageDealt}</td>
+                        <td className="text-right px-3 py-2">{p.avgSpellsCast}</td>
+                        <td className="text-right px-3 py-2">{p.avgCreaturesPlayed}</td>
+                        <td className="text-right px-3 py-2">{p.avgRemovalUsed}</td>
+                      </tr>
+                    );
+                  })}
+                </tbody>
+              </table>
+            </div>
+
+            {/* Per-Game Logs (collapsible) */}
+            {games.length > 0 && (
+              <div className="bg-bg-zone rounded-md border border-border-zone">
+                <button
+                  onClick={() => setShowLogs(!showLogs)}
+                  className="w-full text-left px-4 py-3 text-sm font-medium flex justify-between items-center hover:bg-bg-panel transition-colors"
+                >
+                  <span>Per-Game Results ({games.length} games)</span>
+                  <span className="text-text-muted">{showLogs ? '▲' : '▼'}</span>
+                </button>
+                {showLogs && (
+                  <div className="px-4 pb-3 max-h-96 overflow-y-auto">
+                    <div className="space-y-1">
+                      {games.map((g: unknown, i: number) => {
+                        const game = g as Record<string, unknown>;
+                        return (
+                          <div
+                            key={i}
+                            className="flex items-center gap-3 text-xs text-text-muted py-1 border-b border-border-zone"
+                          >
+                            <span className="w-12 font-mono">#{game.gameNumber as number}</span>
+                            <span>
+                              Winner:{' '}
+                              <span
+                                style={{
+                                  color: SEAT_COLOURS[(game.winner ?? game.winningSeat ?? 0) as number],
+                                }}
+                              >
+                                P{((game.winner ?? game.winningSeat ?? 0) as number) + 1}
+                              </span>
+                            </span>
+                            <span>{game.turns as number} turns</span>
+                          </div>
+                        );
+                      })}
+                    </div>
+                  </div>
+                )}
+              </div>
+            )}
+          </div>
+        )}
+      </div>
+    </div>
+  );
+}

frontend/src/store/gameReducer.ts

@@ -0,0 +1,61 @@
+import type { GameState, Card, NarrationLine } from '../types/game';
+
+export type GameAction =
+  | { type: 'SET_STATE'; payload: GameState }
+  | { type: 'SET_HAND'; cards: Card[] }
+  | { type: 'ADD_NARRATION'; line: NarrationLine }
+  | { type: 'SET_THINKING'; seat: number; active: boolean }
+  | { type: 'GAME_OVER'; winner: number; reason: string }
+  | { type: 'SET_STATUS'; status: string };
+
+export interface ReducerState {
+  gameState: GameState | null;
+  hand: Card[];
+  narration: NarrationLine[];
+  thinkingSeats: Set<number>;
+  gameOver: { winner: number; reason: string } | null;
+  status: string;
+}
+
+export const initialState: ReducerState = {
+  gameState: null,
+  hand: [],
+  narration: [],
+  thinkingSeats: new Set(),
+  gameOver: null,
+  status: '',
+};
+
+export function gameReducer(state: ReducerState, action: GameAction): ReducerState {
+  switch (action.type) {
+    case 'SET_STATE':
+      return { ...state, gameState: action.payload };
+
+    case 'SET_HAND':
+      return { ...state, hand: action.cards };
+
+    case 'ADD_NARRATION': {
+      const narration = [...state.narration, action.line];
+      return { ...state, narration: narration.slice(-120) };
+    }
+
+    case 'SET_THINKING': {
+      const thinkingSeats = new Set(state.thinkingSeats);
+      if (action.active) {
+        thinkingSeats.add(action.seat);
+      } else {
+        thinkingSeats.delete(action.seat);
+      }
+      return { ...state, thinkingSeats };
+    }
+
+    case 'GAME_OVER':
+      return { ...state, gameOver: { winner: action.winner, reason: action.reason } };
+
+    case 'SET_STATUS':
+      return { ...state, status: action.status };
+
+    default:
+      return state;
+  }
+}

frontend/src/test/setup.ts

@@ -0,0 +1 @@
+import '@testing-library/jest-dom';

frontend/src/types/game.ts

@@ -0,0 +1,55 @@
+export type Phase =
+  | 'untap' | 'upkeep' | 'draw' | 'main1'
+  | 'begin_combat' | 'declare_attackers' | 'declare_blockers' | 'combat_damage'
+  | 'main2' | 'end_step' | 'cleanup';
+
+export interface Card {
+  name: string;
+  type: string;
+  tapped: boolean;
+  is_commander: boolean;
+  oracle: string;
+  cmc: number;
+}
+
+export interface Player {
+  seat: number;
+  name: string;
+  life: number;
+  eliminated: boolean;
+  hand: Card[];
+  battlefield: Card[];
+  graveyard: Card[];
+  exile: Card[];
+  command_zone: Card[];
+  hand_count: number;
+}
+
+export interface GameState {
+  game_id: string;
+  turn: number;
+  current_phase: Phase;
+  active_seat: number;
+  players: Player[];
+  stack: Card[];
+  game_over: boolean;
+  winner: Player | null;
+}
+
+export type WsIncoming =
+  | { type: 'state'; payload: GameState }
+  | { type: 'event'; text: string; seat?: number }
+  | { type: 'phase'; phase: Phase; active_seat: number }
+  | { type: 'thinking'; seat: number; active: boolean }
+  | { type: 'elimination'; seat: number }
+  | { type: 'game_over'; winner: number; reason: string }
+  | { type: 'hand'; cards: Card[] }
+  | { type: 'pong' }
+  | { type: 'error'; message: string };
+
+export interface NarrationLine {
+  text: string;
+  seat?: number;
+  isSystem: boolean;
+  timestamp: number;
+}

frontend/src/types/simulator.ts

@@ -0,0 +1,28 @@
+export interface SimDeck {
+  id: number;
+  name: string;
+  commander_name: string;
+  color_identity: string[];
+}
+
+export interface SimPlayerResult {
+  seat: number;
+  deckName: string;
+  wins: number;
+  winRate: number;
+  avgTurns: number;
+  avgDamageDealt: number;
+  avgSpellsCast: number;
+  avgCreaturesPlayed: number;
+  avgRemovalUsed: number;
+  avgRampPlayed: number;
+  avgCardsDrawn: number;
+  avgMaxBoardSize: number;
+}
+
+export interface SimResult {
+  playerCount: number;
+  totalGames: number;
+  players: SimPlayerResult[];
+  elapsedSeconds: number;
+}

frontend/tailwind.config.js

@@ -0,0 +1,82 @@
+/** @type {import('tailwindcss').Config} */
+export default {
+  content: [
+    "./index.html",
+    "./src/**/*.{js,ts,jsx,tsx}",
+  ],
+  theme: {
+    extend: {
+      colors: {
+        bg: {
+          base: '#12141e',
+          zone: '#1c2034',
+          panel: '#161820',
+          bar: '#1a1e30',
+          phase: '#1e2338',
+          hand: '#1a2030',
+        },
+        accent: {
+          blue: '#3c82dc',
+          think: '#64b4ff',
+          purple: '#7060c0',
+        },
+        life: {
+          high: '#50c87a',
+          med: '#dcb432',
+          low: '#dc3c3c',
+        },
+        seat: {
+          0: '#4696ff',
+          1: '#ff6450',
+          2: '#50d278',
+          3: '#dcaa32',
+        },
+        card: {
+          bg: '#2a3050',
+          hand: '#2a3868',
+          selected: '#3a4878',
+        },
+        border: {
+          zone: '#3a4060',
+          sep: '#2a2e48',
+          btn: '#4060a0',
+          overlay: '#4060c0',
+          'btn-hover': '#6080d0',
+          'btn-primary': '#4080d0',
+          'hand-card': '#5060a0',
+          chip: '#444444',
+        },
+        phase: {
+          tab: '#252a42',
+          text: '#888888',
+        },
+        btn: {
+          primary: '#2a5090',
+        },
+        text: {
+          body: '#e8e8e8',
+          muted: '#888888',
+          system: '#555566',
+          'zone-label': '#666666',
+          status: '#777788',
+          sub: '#aaaaaa',
+          btn: '#cccccc',
+        },
+        highlight: {
+          gold: '#ffdc32',
+          commander: '#ffd200',
+        },
+      },
+      keyframes: {
+        pulse: {
+          '0%, 100%': { opacity: '1' },
+          '50%': { opacity: '0.7' },
+        },
+      },
+      animation: {
+        pulse: 'pulse 1s ease-in-out infinite',
+      },
+    },
+  },
+  plugins: [],
+}

frontend/tsconfig.app.json

@@ -0,0 +1,28 @@
+{
+  "compilerOptions": {
+    "tsBuildInfoFile": "./node_modules/.tmp/tsconfig.app.tsbuildinfo",
+    "target": "ES2023",
+    "useDefineForClassFields": true,
+    "lib": ["ES2023", "DOM", "DOM.Iterable"],
+    "module": "ESNext",
+    "types": ["vite/client"],
+    "skipLibCheck": true,
+
+    /* Bundler mode */
+    "moduleResolution": "bundler",
+    "allowImportingTsExtensions": true,
+    "verbatimModuleSyntax": true,
+    "moduleDetection": "force",
+    "noEmit": true,
+    "jsx": "react-jsx",
+
+    /* Linting */
+    "strict": true,
+    "noUnusedLocals": true,
+    "noUnusedParameters": true,
+    "erasableSyntaxOnly": true,
+    "noFallthroughCasesInSwitch": true,
+    "noUncheckedSideEffectImports": true
+  },
+  "include": ["src"]
+}

frontend/tsconfig.json

@@ -0,0 +1,7 @@
+{
+  "files": [],
+  "references": [
+    { "path": "./tsconfig.app.json" },
+    { "path": "./tsconfig.node.json" }
+  ]
+}

frontend/tsconfig.node.json

@@ -0,0 +1,26 @@
+{
+  "compilerOptions": {
+    "tsBuildInfoFile": "./node_modules/.tmp/tsconfig.node.tsbuildinfo",
+    "target": "ES2023",
+    "lib": ["ES2023"],
+    "module": "ESNext",
+    "types": ["node"],
+    "skipLibCheck": true,
+
+    /* Bundler mode */
+    "moduleResolution": "bundler",
+    "allowImportingTsExtensions": true,
+    "verbatimModuleSyntax": true,
+    "moduleDetection": "force",
+    "noEmit": true,
+
+    /* Linting */
+    "strict": true,
+    "noUnusedLocals": true,
+    "noUnusedParameters": true,
+    "erasableSyntaxOnly": true,
+    "noFallthroughCasesInSwitch": true,
+    "noUncheckedSideEffectImports": true
+  },
+  "include": ["vite.config.ts"]
+}

frontend/vite.config.ts

@@ -0,0 +1,21 @@
+import { defineConfig } from 'vite'
+import react from '@vitejs/plugin-react'
+
+// https://vite.dev/config/
+export default defineConfig({
+  plugins: [react()],
+  server: {
+    // Dev-only proxy — in production FastAPI serves the built assets
+    // and handles /api and /ws routes directly.
+    proxy: {
+      '/api': {
+        target: 'http://localhost:8000',
+        changeOrigin: true,
+      },
+      '/ws': {
+        target: 'ws://localhost:8000',
+        ws: true,
+      },
+    },
+  },
+})

frontend/vitest.config.ts

@@ -0,0 +1,12 @@
+/// <reference types="vitest" />
+import { defineConfig } from 'vitest/config'
+import react from '@vitejs/plugin-react'
+
+export default defineConfig({
+  plugins: [react()],
+  test: {
+    globals: true,
+    environment: 'jsdom',
+    setupFiles: './src/test/setup.ts',
+  },
+})

ml/actions/labeler.py

@@ -240,12 +240,12 @@ def print_label_distribution(labels: np.ndarray) -> Dict[str, int]:
         dist[action.value] = count
 
     total = len(labels)
-    print(f"\n{'Action':<25} {'Count':>8} {'Pct':>8}")
-    print("─" * 43)
+    log.info(f"\n{'Action':<25} {'Count':>8} {'Pct':>8}")
+    log.info("─" * 43)
     for action_name, count in sorted(dist.items(), key=lambda x: -x[1]):
         pct = (count / total * 100) if total > 0 else 0
-        print(f"  {action_name:<23} {count:>8} {pct:>7.1f}%")
-    print(f"{'─' * 43}")
-    print(f"  {'TOTAL':<23} {total:>8}")
+        log.info(f"  {action_name:<23} {count:>8} {pct:>7.1f}%")
+    log.info(f"{'─' * 43}")
+    log.info(f"  {'TOTAL':<23} {total:>8}")
 
     return dist

ml/data/dataset_builder.py

@@ -154,17 +154,17 @@ def build_dataset(
 
     # Print game outcome distribution
     unique_outcomes, outcome_counts = np.unique(outcomes, return_counts=True)
-    print(f"\n{'Outcome':<25} {'Count':>8}")
-    print("─" * 35)
+    logger.info(f"\n{'Outcome':<25} {'Count':>8}")
+    logger.info("─" * 35)
     for o, c in sorted(zip(unique_outcomes, outcome_counts), key=lambda x: -x[1]):
-        print(f"  {str(o):<23} {c:>8}")
+        logger.info(f"  {str(o):<23} {c:>8}")
 
     # Print playstyle distribution
     unique_styles, style_counts = np.unique(playstyles, return_counts=True)
-    print(f"\n{'Playstyle':<25} {'Count':>8}")
-    print("─" * 35)
+    logger.info(f"\n{'Playstyle':<25} {'Count':>8}")
+    logger.info("─" * 35)
     for s, c in sorted(zip(unique_styles, style_counts), key=lambda x: -x[1]):
-        print(f"  {str(s):<23} {c:>8}")
+        logger.info(f"  {str(s):<23} {c:>8}")
 
     return {
         "states": states,
@@ -288,7 +288,7 @@ def main():
         max_samples=args.max_samples,
     )
     if not dataset:
-        print("\nNo data to save. Run simulations with --ml-log first.")
+        logger.info("\nNo data to save. Run simulations with --ml-log first.")
         sys.exit(1)
 
     if args.no_split:
@@ -300,7 +300,7 @@ def main():
         save_dataset(val, f"{base}-val.npz")
         save_dataset(test, f"{base}-test.npz")
 
-    print("\nDataset build complete.")
+    logger.info("\nDataset build complete.")
 
 
 if __name__ == "__main__":

ml/scripts/update_weights.py

@@ -0,0 +1,344 @@
+"""
+update_weights.py
+=================
+Outcome-weighted reinforcement learning for Commander AI card weights.
+
+Reads all decisions_*.jsonl files produced by DeepSeekBrain.flush_log(),
+applies a gradient-free reward nudge to each weight key based on the
+normalised reward signal (-1.0 to +1.0) stamped on every decision entry,
+then saves the updated weights to learned_weights.json via save_weights().
+
+Usage
+-----
+    # Basic run (reads logs/decisions/, updates learned_weights.json)
+    python -m ml.scripts.update_weights
+
+    # Custom paths
+    python -m ml.scripts.update_weights \\
+        --log-dir logs/decisions \\
+        --weights-out src/commander_ai_lab/sim/learned_weights.json \\
+        --lr 0.05 \\
+        --min-games 10
+
+    # Dry-run (print proposed changes, don't write)
+    python -m ml.scripts.update_weights --dry-run
+
+    # Reset to defaults then exit
+    python -m ml.scripts.update_weights --reset
+
+Algorithm
+---------
+For each decision entry that has a non-null game_result:
+
+    delta = learning_rate * reward * action_weight_for(action)
+
+Each weight key is nudged by the mean delta across all decisions that
+touched that key.  Keys with no decisions are left unchanged.
+All values are clamped to [-20.0, 20.0] by save_weights().
+
+Reward is the normalised life-delta from flush_log():
+    reward = (my_final_life - avg_opp_final_life) / 40.0  in [-1.0, 1.0]
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import logging
+import os
+import sys
+from collections import defaultdict
+from pathlib import Path
+from typing import Iterator
+
+# ---------------------------------------------------------------------------
+# Allow running as a script without installing the package
+# ---------------------------------------------------------------------------
+_REPO_ROOT = Path(__file__).resolve().parents[2]
+if str(_REPO_ROOT / "src") not in sys.path:
+    sys.path.insert(0, str(_REPO_ROOT / "src"))
+
+from commander_ai_lab.sim.rules import (
+    AI_DEFAULT_WEIGHTS,
+    load_weights,
+    reset_weights,
+    save_weights,
+)
+
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(asctime)s  %(levelname)-8s  %(message)s",
+    datefmt="%H:%M:%S",
+)
+logger = logging.getLogger("update_weights")
+
+# ---------------------------------------------------------------------------
+# Action → weight key mapping
+# Maps each VALID_ACTION to the weight(s) it most directly influences.
+# A single action may touch multiple keys (listed as a tuple).
+# ---------------------------------------------------------------------------
+ACTION_WEIGHT_MAP: dict[str, tuple[str, ...]] = {
+    "play_land":       (),                                    # no weight key — always correct
+    "cast_ramp":       ("spell_ramp",),
+    "cast_removal":    ("spell_destroy",),
+    "cast_board_wipe": ("spell_counter",),                   # reuse counter as board-control proxy
+    "cast_creature":   ("card_ptBonus",),
+    "cast_spell":      ("spell_enchantment", "spell_draw"),
+    "attack_all":      ("kw_haste", "kw_flying"),            # aggression / evasion proxy
+    "attack_safe":     ("kw_flying", "kw_trample"),
+    "hold":            (),                                    # no weight key — passing is neutral
+}
+
+
+# ---------------------------------------------------------------------------
+# JSONL helpers
+# ---------------------------------------------------------------------------
+
+def _iter_log_entries(log_dir: str) -> Iterator[dict]:
+    """Yield every entry from all decisions_*.jsonl files in log_dir."""
+    log_path = Path(log_dir)
+    if not log_path.exists():
+        logger.warning("Log directory not found: %s", log_dir)
+        return
+
+    files = sorted(log_path.glob("decisions_*.jsonl"))
+    if not files:
+        logger.warning("No decisions_*.jsonl files found in %s", log_dir)
+        return
+
+    logger.info("Found %d log file(s) in %s", len(files), log_dir)
+    for fpath in files:
+        with open(fpath, encoding="utf-8") as f:
+            for lineno, line in enumerate(f, 1):
+                line = line.strip()
+                if not line:
+                    continue
+                try:
+                    yield json.loads(line)
+                except json.JSONDecodeError as exc:
+                    logger.warning("Skipping malformed line %d in %s: %s", lineno, fpath.name, exc)
+
+
+# ---------------------------------------------------------------------------
+# Core learning loop
+# ---------------------------------------------------------------------------
+
+def compute_weight_updates(
+    log_dir: str,
+    current_weights: dict[str, float],
+    learning_rate: float = 0.05,
+    min_games: int = 1,
+) -> tuple[dict[str, float], dict]:
+    """
+    Read decision logs and compute updated weights.
+
+    Returns
+    -------
+    updated_weights : dict[str, float]
+        New weight dict after applying mean reward nudges.
+    stats : dict
+        Diagnostic info (games seen, decisions processed, keys changed, etc.)
+    """
+    # Accumulators: key → list of (reward * lr) deltas
+    key_deltas: dict[str, list[float]] = defaultdict(list)
+
+    total_entries = 0
+    labelled_entries = 0  # entries with non-null game_result
+    game_ids_seen: set[str] = set()
+    action_counts: dict[str, int] = defaultdict(int)
+
+    for entry in _iter_log_entries(log_dir):
+        total_entries += 1
+
+        game_result = entry.get("game_result")
+        if not game_result:
+            continue  # no outcome stamped — skip (pre-flush or incomplete game)
+
+        labelled_entries += 1
+        game_ids_seen.add(entry.get("game_id", "unknown"))
+
+        reward: float = float(game_result.get("reward", 0.0))
+        action: str = entry.get("decision", {}).get("action", "hold")
+        action_counts[action] += 1
+
+        weight_keys = ACTION_WEIGHT_MAP.get(action, ())
+        for key in weight_keys:
+            if key in current_weights:
+                key_deltas[key].append(learning_rate * reward)
+
+    num_games = len(game_ids_seen)
+    logger.info(
+        "Processed %d entries | %d labelled | %d unique games",
+        total_entries, labelled_entries, num_games,
+    )
+
+    if num_games < min_games:
+        logger.warning(
+            "Only %d game(s) found — minimum is %d.  Skipping update.",
+            num_games, min_games,
+        )
+        return dict(current_weights), {
+            "skipped": True,
+            "reason": f"need {min_games} games, have {num_games}",
+            "total_entries": total_entries,
+            "labelled_entries": labelled_entries,
+            "games": num_games,
+        }
+
+    # Apply mean delta per key
+    updated = dict(current_weights)
+    changed_keys: dict[str, tuple[float, float]] = {}  # key → (old, new)
+
+    for key, deltas in key_deltas.items():
+        if not deltas:
+            continue
+        mean_delta = sum(deltas) / len(deltas)
+        old_val = updated[key]
+        new_val = round(old_val + mean_delta, 6)
+        updated[key] = new_val
+        if abs(new_val - old_val) >= 0.0001:
+            changed_keys[key] = (old_val, new_val)
+
+    stats = {
+        "skipped": False,
+        "total_entries": total_entries,
+        "labelled_entries": labelled_entries,
+        "games": num_games,
+        "keys_changed": len(changed_keys),
+        "changes": changed_keys,
+        "action_counts": dict(action_counts),
+    }
+    return updated, stats
+
+
+def print_diff(stats: dict, current: dict, updated: dict) -> None:
+    """Pretty-print weight changes to stdout."""
+    changes = stats.get("changes", {})
+    if not changes:
+        print("  (no weight changes above threshold)")
+        return
+
+    col_w = max(len(k) for k in changes) + 2
+    print(f"  {'Key':<{col_w}}  {'Before':>10}  {'After':>10}  {'Delta':>10}")
+    print("  " + "-" * (col_w + 36))
+    for key, (old, new) in sorted(changes.items(), key=lambda x: abs(x[1][1] - x[1][0]), reverse=True):
+        delta = new - old
+        arrow = "▲" if delta > 0 else "▼"
+        print(f"  {key:<{col_w}}  {old:>10.4f}  {new:>10.4f}  {arrow} {abs(delta):>8.4f}")
+
+
+# ---------------------------------------------------------------------------
+# CLI
+# ---------------------------------------------------------------------------
+
+def _default_log_dir() -> str:
+    return str(_REPO_ROOT / "logs" / "decisions")
+
+
+def _default_weights_out() -> str:
+    return str(_REPO_ROOT / "src" / "commander_ai_lab" / "sim" / "learned_weights.json")
+
+
+def main(argv: list[str] | None = None) -> int:
+    parser = argparse.ArgumentParser(
+        description="Update AI card weights from simulation decision logs."
+    )
+    parser.add_argument(
+        "--log-dir",
+        default=_default_log_dir(),
+        help="Directory containing decisions_*.jsonl files (default: logs/decisions/)",
+    )
+    parser.add_argument(
+        "--weights-out",
+        default=_default_weights_out(),
+        help="Path to write learned_weights.json (default: src/.../sim/learned_weights.json)",
+    )
+    parser.add_argument(
+        "--lr",
+        type=float,
+        default=0.05,
+        help="Learning rate — how much each game nudges a weight (default: 0.05)",
+    )
+    parser.add_argument(
+        "--min-games",
+        type=int,
+        default=10,
+        help="Minimum number of completed games required before updating (default: 10)",
+    )
+    parser.add_argument(
+        "--dry-run",
+        action="store_true",
+        help="Print proposed changes without writing learned_weights.json",
+    )
+    parser.add_argument(
+        "--reset",
+        action="store_true",
+        help="Delete learned_weights.json and restore AI_DEFAULT_WEIGHTS, then exit",
+    )
+    parser.add_argument(
+        "--verbose",
+        action="store_true",
+        help="Show action distribution and per-key delta counts",
+    )
+    args = parser.parse_args(argv)
+
+    # ── Reset mode ──────────────────────────────────────────────
+    if args.reset:
+        reset_weights(args.weights_out)
+        print(f"Reset: deleted {args.weights_out} (if it existed). Defaults restored.")
+        return 0
+
+    # ── Load current weights ─────────────────────────────────────
+    current = load_weights(args.weights_out)
+    logger.info("Loaded %d weight keys from %s", len(current), args.weights_out)
+
+    # ── Compute updates ──────────────────────────────────────────
+    updated, stats = compute_weight_updates(
+        log_dir=args.log_dir,
+        current_weights=current,
+        learning_rate=args.lr,
+        min_games=args.min_games,
+    )
+
+    # ── Report ───────────────────────────────────────────────────
+    print()
+    print("=" * 60)
+    print("  Commander AI — Weight Update Report")
+    print("=" * 60)
+    print(f"  Log dir    : {args.log_dir}")
+    print(f"  Weights out: {args.weights_out}")
+    print(f"  Games      : {stats['games']}")
+    print(f"  Entries    : {stats['total_entries']}  ({stats['labelled_entries']} labelled)")
+    print(f"  LR         : {args.lr}")
+    print()
+
+    if stats.get("skipped"):
+        print(f"  SKIPPED — {stats['reason']}")
+        print()
+        return 0
+
+    print(f"  Keys changed: {stats['keys_changed']}")
+    print()
+    print_diff(stats, current, updated)
+    print()
+
+    if args.verbose:
+        print("  Action distribution:")
+        for action, count in sorted(stats["action_counts"].items(), key=lambda x: -x[1]):
+            print(f"    {action:<20} {count:>6} decisions")
+        print()
+
+    # ── Write ────────────────────────────────────────────────────
+    if args.dry_run:
+        print("  DRY RUN — no file written.")
+    else:
+        out_path = save_weights(updated, args.weights_out)
+        print(f"  Saved updated weights to: {out_path}")
+
+    print("=" * 60)
+    print()
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

ml/training/policy_network.py

@@ -16,6 +16,10 @@ The network is designed for:
 import os
 import sys
 from pathlib import Path
+import logging
+
+log = logging.getLogger("commander_ai_lab.ml.training")
+
 
 # Conditional torch import — graceful failure if not installed
 try:
@@ -25,7 +29,7 @@ try:
     TORCH_AVAILABLE = True
 except ImportError:
     TORCH_AVAILABLE = False
-    print("[ML] PyTorch not installed. Install with: pip install torch")
+    log.warning("PyTorch not installed. Install with: pip install torch")
 
 import numpy as np
 
@@ -204,7 +208,7 @@ def save_checkpoint(
             "dropout": TRAINING_CONFIG.dropout,
         },
     }, path)
-    print(f"[ML] Checkpoint saved: {path}")
+    log.info(f"Checkpoint saved: {path}")
 
 
 def load_checkpoint(path: str, device: str = "cpu"):

ml/training/trainer.py

@@ -49,8 +49,7 @@ from ml.training.policy_network import (
     PolicyNetwork, save_checkpoint, load_checkpoint,
 )
 
-logging.basicConfig(level=logging.INFO, format="[%(name)s] %(message)s")
-logger = logging.getLogger("ml.trainer")
+logger = logging.getLogger("commander_ai_lab.ml.training")
 
 
 # ══════════════════════════════════════════════════════════
@@ -502,40 +501,40 @@ def evaluate_model(
 
 def print_evaluation(results: Dict):
     """Pretty-print evaluation results."""
-    print()
-    print("=" * 65)
-    print("  Test Set Evaluation")
-    print("=" * 65)
-    print(f"  Overall Accuracy: {results['overall_accuracy']:.1%}")
-    print(f"  Total Samples:    {results['total_samples']:,}")
-    print()
-    print(f"  {'Action':<20} {'Support':>8} {'Prec':>7} {'Recall':>7} {'F1':>7}")
-    print("  " + "-" * 55)
+    logger.info("")
+    logger.info("=" * 65)
+    logger.info("  Test Set Evaluation")
+    logger.info("=" * 65)
+    logger.info(f"  Overall Accuracy: {results['overall_accuracy']:.1%}")
+    logger.info(f"  Total Samples:    {results['total_samples']:,}")
+    logger.info("")
+    logger.info(f"  {'Action':<20} {'Support':>8} {'Prec':>7} {'Recall':>7} {'F1':>7}")
+    logger.info("  " + "-" * 55)
 
     for action_name, m in results["per_class"].items():
-        print(
+        logger.info(
             f"  {action_name:<20} {m['support']:>8} "
             f"{m['precision']:>7.3f} {m['recall']:>7.3f} {m['f1']:>7.3f}"
         )
 
-    print("  " + "-" * 55)
+    logger.info("  " + "-" * 55)
 
     # Confusion matrix
-    print()
-    print("  Confusion Matrix (rows=true, cols=predicted):")
+    logger.info("")
+    logger.info("  Confusion Matrix (rows=true, cols=predicted):")
     header = "  " + " " * 18
     for c in range(NUM_ACTIONS):
         short_name = IDX_TO_ACTION[c].value[:6]
         header += f" {short_name:>6}"
-    print(header)
+    logger.info(header)
 
     for r in range(NUM_ACTIONS):
         row_name = IDX_TO_ACTION[r].value[:16]
         row_str = f"  {row_name:<18}"
         for c in range(NUM_ACTIONS):
             row_str += f" {results['confusion_matrix'][r][c]:>6}"
-        print(row_str)
-    print()
+        logger.info(row_str)
+    logger.info("")
 
 
 # ══════════════════════════════════════════════════════════
@@ -605,8 +604,8 @@ def main():
     args = parser.parse_args()
 
     if not TORCH_AVAILABLE:
-        print("PyTorch not installed. Install with:")
-        print("  pip install torch torchvision --index-url https://download.pytorch.org/whl/cu121")
+        logger.info("PyTorch not installed. Install with:")
+        logger.info("  pip install torch torchvision --index-url https://download.pytorch.org/whl/cu121")
         sys.exit(1)
 
     device = args.device or get_best_device()
@@ -621,13 +620,13 @@ def main():
     if args.eval_only:
         # Evaluation only
         if not os.path.exists(test_path):
-            print(f"Test data not found: {test_path}")
-            print("Run: python -m ml.scripts.ml_cli build")
+            logger.info(f"Test data not found: {test_path}")
+            logger.info("Run: python -m ml.scripts.ml_cli build")
             sys.exit(1)
 
         ckpt_path = args.checkpoint or os.path.join(ckpt_dir, "best_policy.pt")
         if not os.path.exists(ckpt_path):
-            print(f"Checkpoint not found: {ckpt_path}")
+            logger.info(f"Checkpoint not found: {ckpt_path}")
             sys.exit(1)
 
         test_states, test_labels = load_npz_split(test_path)
@@ -641,15 +640,15 @@ def main():
         eval_path = os.path.join(ckpt_dir, "eval_results.json")
         with open(eval_path, "w") as f:
             json.dump(results, f, indent=2)
-        print(f"  Results saved: {eval_path}")
+        logger.info(f"  Results saved: {eval_path}")
         return
 
     # Training mode
     for path, label in [(train_path, "train"), (val_path, "val")]:
         if not os.path.exists(path):
-            print(f"{label} data not found: {path}")
-            print("Build the dataset first:")
-            print("  python -m ml.scripts.ml_cli build --results-dir results/")
+            logger.info(f"{label} data not found: {path}")
+            logger.info("Build the dataset first:")
+            logger.info("  python -m ml.scripts.ml_cli build --results-dir results/")
             sys.exit(1)
 
     train_states, train_labels = load_npz_split(train_path)

models/__init__.py

@@ -0,0 +1,47 @@
+"""Pydantic models for Commander AI Lab API."""
+
+from models.requests import (
+    StartRequest,
+    ImportUrlRequest,
+    ImportTextRequest,
+    MetaFetchRequest,
+    CreateDeckRequest,
+    UpdateDeckRequest,
+    AddDeckCardRequest,
+    PatchDeckCardRequest,
+    BulkAddRequest,
+    BulkAddRecommendedRequest,
+    DeckGenerationSourceConfig,
+    DeckGenerationRequest,
+    DeckGenV3Request,
+    DeckGenV3SubstituteRequest,
+)
+from models.responses import (
+    StartResponse,
+    StatusResponse,
+    DeckInfo,
+    GeneratedDeckCard,
+)
+
+__all__ = [
+    # Requests
+    "StartRequest",
+    "ImportUrlRequest",
+    "ImportTextRequest",
+    "MetaFetchRequest",
+    "CreateDeckRequest",
+    "UpdateDeckRequest",
+    "AddDeckCardRequest",
+    "PatchDeckCardRequest",
+    "BulkAddRequest",
+    "BulkAddRecommendedRequest",
+    "DeckGenerationSourceConfig",
+    "DeckGenerationRequest",
+    "DeckGenV3Request",
+    "DeckGenV3SubstituteRequest",
+    # Responses
+    "StartResponse",
+    "StatusResponse",
+    "DeckInfo",
+    "GeneratedDeckCard",
+]

models/requests.py

@@ -0,0 +1,175 @@
+"""Pydantic request models for Commander AI Lab API."""
+from __future__ import annotations
+from typing import Optional
+from pydantic import BaseModel
+
+
+class StartRequest(BaseModel):
+    decks: list[str]
+    numGames: int = 100
+    threads: int = 4
+    seed: Optional[int] = None
+    clock: int = 300
+    deckSources: Optional[list[Optional[dict]]] = None
+    useLearnedPolicy: bool = False
+    policyStyle: str = "midrange"
+    policyGreedy: bool = False
+    aiSimplified: bool = False
+    aiThinkTimeMs: int = -1
+    maxQueueDepth: int = -1
+
+
+class ImportUrlRequest(BaseModel):
+    url: str
+
+
+class ImportTextRequest(BaseModel):
+    text: str
+    commander: Optional[str] = None
+
+
+class MetaFetchRequest(BaseModel):
+    commander: str
+
+
+class CreateDeckRequest(BaseModel):
+    name: str
+    commander_scryfall_id: Optional[str] = ""
+    commander_name: Optional[str] = ""
+    color_identity: Optional[list] = []
+    strategy_tag: Optional[str] = ""
+
+
+class UpdateDeckRequest(BaseModel):
+    name: Optional[str] = None
+    commander_scryfall_id: Optional[str] = None
+    commander_name: Optional[str] = None
+    color_identity: Optional[list] = None
+    strategy_tag: Optional[str] = None
+
+
+class AddDeckCardRequest(BaseModel):
+    scryfall_id: str
+    card_name: Optional[str] = ""
+    quantity: Optional[int] = 1
+    is_commander: Optional[int] = 0
+    role_tag: Optional[str] = ""
+
+
+class PatchDeckCardRequest(BaseModel):
+    quantity: Optional[int] = None
+    role_tag: Optional[str] = None
+
+
+class BulkAddRequest(BaseModel):
+    cards: list[dict]
+    respect_ratios: Optional[bool] = False
+
+
+class BulkAddRecommendedRequest(BaseModel):
+    source: str = "collection"
+    only_owned: Optional[bool] = True
+    respect_ratios: Optional[bool] = False
+    types: Optional[list[str]] = None
+    roles: Optional[list[str]] = None
+
+
+class DeckGenerationSourceConfig(BaseModel):
+    use_archidekt: bool = True
+    use_edhrec: bool = True
+    use_moxfield: bool = False
+    use_mtggoldfish: bool = False
+    archidekt_url: Optional[str] = ""
+    moxfield_url: Optional[str] = ""
+    mtggoldfish_url: Optional[str] = ""
+
+
+class DeckGenerationRequest(BaseModel):
+    commander_name: Optional[str] = ""
+    commander_scryfall_id: Optional[str] = ""
+    color_identity: Optional[list[str]] = None
+    sources: Optional[DeckGenerationSourceConfig] = None
+    target_land_count: int = 37
+    target_instant_count: int = 10
+    target_sorcery_count: int = 8
+    target_artifact_count: int = 10
+    target_enchantment_count: int = 8
+    target_creature_count: int = 25
+    target_planeswalker_count: int = 2
+    only_cards_in_collection: bool = False
+    allow_proxies: bool = True
+    deck_name: Optional[str] = ""
+
+
+class DeckGenV3Request(BaseModel):
+    commander_name: str = ""
+    strategy: str = ""
+    target_bracket: int = 3
+    budget_usd: Optional[float] = None
+    budget_mode: str = "total"
+    omit_cards: list[str] = []
+    use_collection: bool = True
+    run_substitution: bool = True
+    model: Optional[str] = None
+    deck_name: Optional[str] = ""
+
+
+
+
+# ══════════════════════════════════════════════════════════════
+# Deckgen requests (moved from routes/deckgen.py)
+# ══════════════════════════════════════════════════════════════
+
+
+class DeckResearchRequest(BaseModel):
+    deck_id: int  # Deck to research
+    goal: Optional[str] = "Identify weaknesses and suggest upgrades"
+    budget_usd: Optional[float] = None
+    omit_cards: Optional[list[str]] = []
+    use_collection: bool = True
+
+
+class DeckGenerateAIRequest(BaseModel):
+    commander: str
+    budget_usd: Optional[float] = None
+    budget_mode: str = "total"  # "total" or "per_card"
+    omit_cards: Optional[list[str]] = []
+    use_collection: bool = True
+
+
+# ══════════════════════════════════════════════════════════════
+# Coach requests (moved from routes/coach.py)
+# ══════════════════════════════════════════════════════════════
+
+
+class CoachRequestBody(BaseModel):
+    goals: Optional[dict] = None
+
+
+class CoachChatMessage(BaseModel):
+    role: str  # "user" or "assistant"
+    content: str
+
+
+class CoachChatRequest(BaseModel):
+    deck_id: str
+    messages: list[dict]  # conversation history [{role, content}]
+    goals: Optional[dict] = None
+    stream: Optional[bool] = False
+
+
+class CoachApplyRequest(BaseModel):
+    session_id: str
+    deck_id: int  # numeric deck ID in the DB
+    accepted_cuts: list[str] = []  # card names to remove
+    accepted_adds: list[str] = []  # card names to add
+
+
+class CoachGoalsRequest(BaseModel):
+    target_power_level: Optional[int] = None  # 1-10
+    meta_focus: Optional[str] = None  # aggro, control, combo, midrange, stax
+    budget: Optional[str] = None  # budget, medium, no-limit
+    focus_areas: list[str] = []  # e.g., ["ramp", "card draw"]
+class DeckGenV3SubstituteRequest(BaseModel):
+    card_name: str
+    substitute_name: str

models/responses.py

@@ -0,0 +1,45 @@
+"""Pydantic response models for Commander AI Lab API."""
+from __future__ import annotations
+from typing import Optional
+from pydantic import BaseModel
+
+
+class StartResponse(BaseModel):
+    batchId: str
+    status: str = "started"
+    message: str = ""
+
+
+class StatusResponse(BaseModel):
+    batchId: str = ""
+    running: bool = False
+    completed: int = 0
+    total: int = 0
+    threads: int = 0
+    elapsedMs: int = 0
+    error: Optional[str] = None
+    simsPerSec: float = 0.0
+    run_id: str = ""
+    games_completed: int = 0
+    total_games: int = 0
+    current_decks: list = []
+
+
+class DeckInfo(BaseModel):
+    name: str
+    filename: str
+
+
+class GeneratedDeckCard(BaseModel):
+    scryfall_id: str = ""
+    name: str = ""
+    type_line: str = ""
+    mana_cost: str = ""
+    cmc: float = 0
+    card_type: str = ""
+    roles: list[str] = []
+    source: str = "collection"
+    quantity: int = 1
+    image_url: str = ""
+    owned_qty: int = 0
+    is_proxy: bool = False

models/state.py

@@ -0,0 +1,101 @@
+"""Runtime configuration, state classes, and static data for Commander AI Lab."""
+from __future__ import annotations
+import json
+import logging
+import subprocess
+from datetime import datetime
+from pathlib import Path
+from typing import Optional
+
+
+log = logging.getLogger("commander_ai_lab.api")
+
+
+# ══════════════════════════════════════════════════════════════
+# Configuration
+# ══════════════════════════════════════════════════════════════
+class Config:
+    """Runtime configuration -- set via CLI args or env vars."""
+    forge_jar: str = ""
+    forge_dir: str = ""
+    forge_decks_dir: str = ""
+    lab_jar: str = ""
+    precon_dir: str = ""
+    results_dir: str = "results"
+    port: int = 8080
+    ximilar_api_key: str = ""
+    pplx_api_key: str = ""
+
+
+CFG = Config()
+
+
+# ══════════════════════════════════════════════════════════════
+# In-Memory State
+# ══════════════════════════════════════════════════════════════
+class BatchState:
+    def __init__(self, batch_id: str, total_games: int, threads: int):
+        self.batch_id = batch_id
+        self.total_games = total_games
+        self.threads = threads
+        self.completed_games = 0
+        self.running = True
+        self.start_time = datetime.now()
+        self.elapsed_ms = 0
+        self.result_path: Optional[str] = None
+        self.error: Optional[str] = None
+        self.process: Optional[subprocess.Popen] = None
+        self.log_lines: list = []
+        self.sims_per_sec: float = 0.0
+
+
+active_batches: dict[str, BatchState] = {}
+
+
+COMMANDER_META: dict = {}
+
+BUILTIN_COMMANDERS: dict = {
+    "Edgar Markov": [{"source": "edhrec", "archetype": "aggro", "colorIdentity": ["W","B","R"]}],
+    "Atraxa, Praetors' Voice": [{"source": "edhrec", "archetype": "midrange", "colorIdentity": ["W","U","B","G"]}],
+    "Korvold, Fae-Cursed King": [{"source": "edhrec", "archetype": "combo", "colorIdentity": ["B","R","G"]}],
+    "Muldrotha, the Gravetide": [{"source": "edhrec", "archetype": "midrange", "colorIdentity": ["U","B","G"]}],
+    "The Ur-Dragon": [{"source": "edhrec", "archetype": "midrange", "colorIdentity": ["W","U","B","R","G"]}],
+    "Yuriko, the Tiger's Shadow": [{"source": "edhrec", "archetype": "aggro", "colorIdentity": ["U","B"]}],
+    "Krenko, Mob Boss": [{"source": "edhrec", "archetype": "aggro", "colorIdentity": ["R"]}],
+    "Meren of Clan Nel Toth": [{"source": "edhrec", "archetype": "midrange", "colorIdentity": ["B","G"]}],
+    "Prossh, Skyraider of Kher": [{"source": "edhrec", "archetype": "combo", "colorIdentity": ["B","R","G"]}],
+    "Kaalia of the Vast": [{"source": "edhrec", "archetype": "aggro", "colorIdentity": ["W","B","R"]}],
+    "Talrand, Sky Summoner": [{"source": "edhrec", "archetype": "control", "colorIdentity": ["U"]}],
+    "Omnath, Locus of Creation": [{"source": "edhrec", "archetype": "combo", "colorIdentity": ["W","U","R","G"]}],
+    "Teysa Karlov": [{"source": "edhrec", "archetype": "combo", "colorIdentity": ["W","B"]}],
+    "Lathril, Blade of the Elves": [{"source": "edhrec", "archetype": "aggro", "colorIdentity": ["B","G"]}],
+    "Breya, Etherium Shaper": [{"source": "edhrec", "archetype": "combo", "colorIdentity": ["W","U","B","R"]}],
+}
+
+
+def load_commander_meta() -> None:
+    """Load commander meta mapping from file or fall back to builtins."""
+    global COMMANDER_META
+    meta_path = Path(__file__).parent.parent / "commander-meta.json"
+    if meta_path.exists():
+        try:
+            with open(meta_path, "r", encoding="utf-8") as f:
+                COMMANDER_META = json.load(f)
+            log.info(f" Meta: Loaded {len(COMMANDER_META)} commanders from {meta_path}")
+            return
+        except Exception as e:
+            log.warning(f" WARNING: Failed to load commander-meta.json: {e}")
+    COMMANDER_META = BUILTIN_COMMANDERS
+    log.info(f" Meta:    {len(COMMANDER_META)} built-in commanders")
+
+
+# ══════════════════════════════════════════════════════════════
+# AI Profiles
+# ══════════════════════════════════════════════════════════════
+AI_PROFILES = {
+    "default": {"name": "default", "description": "Balanced \u2014 Forge's default AI behavior", "aggression": 0.5, "cardAdvantage": 0.5, "removalPriority": 0.5, "boardPresence": 0.5, "comboPriority": 0.3, "patience": 0.5},
+    "aggro": {"name": "aggro", "description": "Aggressive \u2014 attacks early, prioritizes damage", "aggression": 0.9, "cardAdvantage": 0.3, "removalPriority": 0.3, "boardPresence": 0.8, "comboPriority": 0.1, "patience": 0.1},
+    "control": {"name": "control", "description": "Control \u2014 defensive, removal-heavy, card advantage", "aggression": 0.2, "cardAdvantage": 0.9, "removalPriority": 0.9, "boardPresence": 0.3, "comboPriority": 0.4, "patience": 0.9},
+    "combo": {"name": "combo", "description": "Combo \u2014 ramps, digs for pieces, assembles combos", "aggression": 0.2, "cardAdvantage": 0.8, "removalPriority": 0.4, "boardPresence": 0.3, "comboPriority": 0.95, "patience": 0.7},
+    "midrange": {"name": "midrange", "description": "Midrange \u2014 flexible, strong board presence, value-oriented", "aggression": 0.5, "cardAdvantage": 0.6, "removalPriority": 0.6, "boardPresence": 0.7, "comboPriority": 0.3, "patience": 0.5},
+}

overnight-run.py

@@ -22,11 +22,15 @@ Configuration:
 
 import argparse
 import json
+import logging
+import logging.handlers
+import os
 import sys
 import time
 import urllib.request
 import urllib.error
 from datetime import datetime, timedelta
+from pathlib import Path
 
 
 # ── Default Config ──────────────────────────────────────────────
@@ -51,10 +55,48 @@ DEFAULT_CONFIG = {
 }
 
 
-def log(msg):
-    ts = datetime.now().strftime("%H:%M:%S")
-    print(f"[{ts}] {msg}")
-    sys.stdout.flush()
+# ── Logging ──────────────────────────────────────────────────
+_LOG_DIR = Path(os.environ.get("CAL_LOG_DIR", "logs"))
+
+
+def setup_overnight_logging() -> logging.Logger:
+    """Configure logging for the overnight runner.
+
+    Outputs to both console and a rotating log file so that
+    unattended runs always have a persistent record.
+    """
+    _LOG_DIR.mkdir(parents=True, exist_ok=True)
+
+    logger = logging.getLogger("commander_ai_lab.overnight")
+    logger.setLevel(logging.INFO)
+
+    if logger.handlers:
+        return logger
+
+    fmt = logging.Formatter(
+        "%(asctime)s %(levelname)-5s [%(name)s] %(message)s",
+        datefmt="%Y-%m-%d %H:%M:%S",
+    )
+
+    console = logging.StreamHandler(sys.stdout)
+    console.setLevel(logging.INFO)
+    console.setFormatter(fmt)
+    logger.addHandler(console)
+
+    file_handler = logging.handlers.RotatingFileHandler(
+        _LOG_DIR / "overnight-run.log",
+        maxBytes=10 * 1024 * 1024,  # 10 MB
+        backupCount=5,
+        encoding="utf-8",
+    )
+    file_handler.setLevel(logging.INFO)
+    file_handler.setFormatter(fmt)
+    logger.addHandler(file_handler)
+
+    return logger
+
+
+log = setup_overnight_logging()
 
 
 def api_get(base_url, path):
@@ -129,8 +171,8 @@ def wait_for_batch(base_url, batch_id, engine="deepseek", deadline=None):
     while True:
         # Hard deadline check — stop waiting if we've run out of time
         if deadline and time.time() >= deadline:
-            log(f"  Time limit reached while batch {batch_id} was running.")
-            log(f"  The batch will continue on the server but we're moving on.")
+            log.info(f"  Time limit reached while batch {batch_id} was running.")
+            log.info(f"  The batch will continue on the server but we're moving on.")
             return {"completed": 0, "timed_out": True}
 
         try:
@@ -144,12 +186,12 @@ def wait_for_batch(base_url, batch_id, engine="deepseek", deadline=None):
             elapsed = status.get("elapsedMs", 0)
 
             if error:
-                log(f"  Batch {batch_id} ERROR: {error}")
+                log.error(f"  Batch {batch_id} ERROR: {error}")
                 return None
 
             if not running and completed > 0:
                 elapsed_s = elapsed / 1000 if elapsed else 0
-                log(f"  Batch {batch_id} complete: {completed} games in {elapsed_s:.0f}s")
+                log.info(f"  Batch {batch_id} complete: {completed} games in {elapsed_s:.0f}s")
                 try:
                     result = api_get(base_url, f"/api/lab/result?batchId={batch_id}")
                     return result
@@ -160,15 +202,15 @@ def wait_for_batch(base_url, batch_id, engine="deepseek", deadline=None):
             if not running and completed == 0:
                 stall_checks += 1
                 if stall_checks > 20:  # 5 min with no progress
-                    log(f"  Batch {batch_id} never started. Aborting.")
+                    log.info(f"  Batch {batch_id} never started. Aborting.")
                     return None
 
             # Progress log with ETA
             if deadline:
                 mins_left = (deadline - time.time()) / 60
-                log(f"  Progress: {completed}/{total} games ({elapsed / 1000:.0f}s) — {mins_left:.0f}min left in run")
+                log.info(f"  Progress: {completed}/{total} games ({elapsed / 1000:.0f}s) — {mins_left:.0f}min left in run")
             else:
-                log(f"  Progress: {completed}/{total} games ({elapsed / 1000:.0f}s)")
+                log.info(f"  Progress: {completed}/{total} games ({elapsed / 1000:.0f}s)")
 
             # Track stall
             if completed == last_completed:
@@ -180,9 +222,9 @@ def wait_for_batch(base_url, batch_id, engine="deepseek", deadline=None):
         except Exception as e:
             consecutive_errors += 1
             if consecutive_errors > 10:
-                log(f"  Lost connection to server after 10 retries. Aborting batch.")
+                log.info(f"  Lost connection to server after 10 retries. Aborting batch.")
                 return None
-            log(f"  Connection error (retry {consecutive_errors}/10): {e}")
+            log.info(f"  Connection error (retry {consecutive_errors}/10): {e}")
 
         time.sleep(poll_interval)
 
@@ -214,26 +256,26 @@ def wait_for_training(base_url):
             error = status.get("error")
 
             if error:
-                log(f"  Training ERROR: {error}")
+                log.error(f"  Training ERROR: {error}")
                 return False
 
             if not running and phase in ("done", "idle") and epoch > 0:
                 result = status.get("result", {})
-                log(f"  Training complete!")
+                log.info(f"  Training complete!")
                 if result:
-                    log(f"  Final accuracy: {result.get('best_val_accuracy', '?')}")
-                    log(f"  Model saved: {result.get('model_path', '?')}")
+                    log.info(f"  Final accuracy: {result.get('best_val_accuracy', '?')}")
+                    log.info(f"  Model saved: {result.get('model_path', '?')}")
                 return True
 
             if phase == "training" and total > 0:
                 metrics = status.get("metrics", {})
                 val_acc = metrics.get("val_accuracy", "?") if metrics else "?"
-                log(f"  Training: epoch {epoch}/{total} — {message} (val_acc: {val_acc})")
+                log.info(f"  Training: epoch {epoch}/{total} — {message} (val_acc: {val_acc})")
             else:
-                log(f"  [{phase}] {message}")
+                log.info(f"  [{phase}] {message}")
 
         except Exception as e:
-            log(f"  Connection error: {e}")
+            log.info(f"  Connection error: {e}")
 
         time.sleep(poll_interval)
 
@@ -253,46 +295,46 @@ def run_overnight(cfg):
     """Main overnight pipeline."""
     base_url = cfg["lab_url"]
 
-    log("=" * 60)
-    log("  Commander AI Lab — Overnight AFK Runner")
-    log("=" * 60)
+    log.info("=" * 60)
+    log.info("  Commander AI Lab — Overnight AFK Runner")
+    log.info("=" * 60)
 
     # ── Verify server ────────────────────────────────────────
-    log("Checking lab server...")
+    log.info("Checking lab server...")
     if not check_server(base_url):
-        log(f"ERROR: Cannot reach lab server at {base_url}")
-        log("Start the lab first: start-lab.bat")
+        log.error(f"ERROR: Cannot reach lab server at {base_url}")
+        log.info("Start the lab first: start-lab.bat")
         return False
 
-    log(f"Server OK at {base_url}")
+    log.info(f"Server OK at {base_url}")
 
     # ── Get decks ─────────────────────────────────────────────
     available_decks = get_available_decks(base_url)
-    log(f"Found {len(available_decks)} decks: {', '.join(available_decks[:5])}{'...' if len(available_decks) > 5 else ''}")
+    log.info(f"Found {len(available_decks)} decks: {', '.join(available_decks[:5])}{'...' if len(available_decks) > 5 else ''}")
 
     engine = cfg["engine"]
     if engine == "deepseek":
         decks = cfg["ds_decks"] if cfg["ds_decks"] else available_decks
         if not decks:
-            log("ERROR: No decks available. Create decks in the Deck Builder first.")
+            log.error("ERROR: No decks available. Create decks in the Deck Builder first.")
             return False
-        log(f"Using DeepSeek engine with {len(decks)} deck(s)")
+        log.info(f"Using DeepSeek engine with {len(decks)} deck(s)")
     else:
         decks = cfg["java_decks"]
         if len(decks) != 3:
-            log("ERROR: Java engine requires exactly 3 deck names.")
-            log("Set java_decks in the config or use --decks deck1 deck2 deck3")
+            log.error("ERROR: Java engine requires exactly 3 deck names.")
+            log.info("Set java_decks in the config or use --decks deck1 deck2 deck3")
             return False
-        log(f"Using Java/Forge engine with decks: {decks}")
+        log.info(f"Using Java/Forge engine with decks: {decks}")
 
     # ── ML logging ────────────────────────────────────────────
-    log("Enabling ML logging...")
+    log.info("Enabling ML logging...")
     try:
         api_post(base_url, "/api/ml/toggle?enable=true", {})
-        log("ML logging enabled")
+        log.info("ML logging enabled")
     except Exception as e:
-        log(f"WARNING: Could not enable ML logging: {e}")
-        log("(DeepSeek batches always log ML data regardless)")
+        log.warning(f"WARNING: Could not enable ML logging: {e}")
+        log.info("(DeepSeek batches always log ML data regardless)")
         # Try alternate form
         try:
             api_get(base_url, "/api/ml/toggle?enable=true")
@@ -301,7 +343,7 @@ def run_overnight(cfg):
 
     # ── Initial data stats ────────────────────────────────────
     decisions_before, files_before = get_ml_data_summary(base_url)
-    log(f"Current ML data: {decisions_before} decisions in {files_before} files")
+    log.info(f"Current ML data: {decisions_before} decisions in {files_before} files")
 
     # ── Run batches ───────────────────────────────────────────
     games_per_batch = cfg["games_per_batch"]
@@ -315,32 +357,32 @@ def run_overnight(cfg):
     total_games = 0
     total_decisions = 0
 
-    log("")
-    log(f"Starting batch loop:")
+    log.info("")
+    log.info(f"Starting batch loop:")
     if num_batches > 0:
-        log(f"  Mode: {num_batches} batches × {games_per_batch} games")
+        log.info(f"  Mode: {num_batches} batches × {games_per_batch} games")
     else:
-        log(f"  Mode: Run for {cfg['time_limit_hours']}h, {games_per_batch} games/batch")
-    log(f"  Engine: {engine}")
-    log(f"  Auto-train after: {'Yes' if cfg['auto_train'] else 'No'}")
-    log("")
+        log.info(f"  Mode: Run for {cfg['time_limit_hours']}h, {games_per_batch} games/batch")
+    log.info(f"  Engine: {engine}")
+    log.info(f"  Auto-train after: {'Yes' if cfg['auto_train'] else 'No'}")
+    log.info("")
 
     while True:
         # Check stopping conditions
         if num_batches > 0 and batch_num >= num_batches:
-            log(f"Reached target of {num_batches} batches. Stopping.")
+            log.info(f"Reached target of {num_batches} batches. Stopping.")
             break
 
         elapsed = time.time() - start_time
         remaining = deadline - time.time()
         if remaining < 120:  # need at least 2 min for a batch
-            log(f"Time limit approaching ({remaining:.0f}s left). Stopping sim loop.")
+            log.info(f"Time limit approaching ({remaining:.0f}s left). Stopping sim loop.")
             break
 
         batch_num += 1
         elapsed_str = str(timedelta(seconds=int(elapsed)))
         remaining_str = str(timedelta(seconds=int(remaining)))
-        log(f"── Batch {batch_num} ── (elapsed: {elapsed_str}, remaining: {remaining_str})")
+        log.info(f"── Batch {batch_num} ── (elapsed: {elapsed_str}, remaining: {remaining_str})")
 
         try:
             if engine == "deepseek":
@@ -352,12 +394,12 @@ def run_overnight(cfg):
                 )
 
             if not batch_id:
-                log("  ERROR: Failed to start batch (no batchId returned)")
-                log("  Waiting 60s before retry...")
+                log.error("  ERROR: Failed to start batch (no batchId returned)")
+                log.info("  Waiting 60s before retry...")
                 time.sleep(60)
                 continue
 
-            log(f"  Started batch {batch_id}")
+            log.info(f"  Started batch {batch_id}")
             result = wait_for_batch(base_url, batch_id, engine, deadline=deadline)
 
             if result:
@@ -366,19 +408,19 @@ def run_overnight(cfg):
                 for deck in result.get("decks", []):
                     wr = deck.get("winRate", "?")
                     name = deck.get("deckName", "?")
-                    log(f"    {name}: {wr}% WR")
+                    log.info(f"    {name}: {wr}% WR")
             else:
-                log("  Batch failed, continuing to next...")
+                log.info("  Batch failed, continuing to next...")
 
         except Exception as e:
-            log(f"  ERROR starting batch: {e}")
-            log("  Waiting 60s before retry...")
+            log.error(f"  ERROR starting batch: {e}")
+            log.info("  Waiting 60s before retry...")
             time.sleep(60)
             continue
 
         # Pause between batches
         if pause_sec > 0:
-            log(f"  Cooling down {pause_sec}s...")
+            log.info(f"  Cooling down {pause_sec}s...")
             time.sleep(pause_sec)
 
     # ── Summary ───────────────────────────────────────────────
@@ -387,26 +429,26 @@ def run_overnight(cfg):
     new_decisions = decisions_after - decisions_before
     new_files = files_after - files_before
 
-    log("")
-    log("=" * 60)
-    log("  BATCH SIMULATION COMPLETE")
-    log("=" * 60)
-    log(f"  Batches run:       {batch_num}")
-    log(f"  Total games:       ~{total_games}")
-    log(f"  New ML decisions:  {new_decisions} (in {new_files} new files)")
-    log(f"  Total ML data:     {decisions_after} decisions")
-    log(f"  Elapsed time:      {str(timedelta(seconds=int(total_elapsed)))}")
-    log("")
+    log.info("")
+    log.info("=" * 60)
+    log.info("  BATCH SIMULATION COMPLETE")
+    log.info("=" * 60)
+    log.info(f"  Batches run:       {batch_num}")
+    log.info(f"  Total games:       ~{total_games}")
+    log.info(f"  New ML decisions:  {new_decisions} (in {new_files} new files)")
+    log.info(f"  Total ML data:     {decisions_after} decisions")
+    log.info(f"  Elapsed time:      {str(timedelta(seconds=int(total_elapsed)))}")
+    log.info("")
 
     # ── Auto-train ────────────────────────────────────────────
     if cfg["auto_train"]:
         if decisions_after < 100:
-            log("Skipping training — too few decisions (need at least 100)")
+            log.info("Skipping training — too few decisions (need at least 100)")
             return True
 
-        log("Starting ML training pipeline...")
-        log(f"  Epochs: {cfg['train_epochs']}, LR: {cfg['train_lr']}, "
-            f"Batch: {cfg['train_batch_size']}, Patience: {cfg['train_patience']}")
+        log.info("Starting ML training pipeline...")
+        log.info(f"  Epochs: {cfg['train_epochs']}, LR: {cfg['train_lr']}, "
+                 f"Batch: {cfg['train_batch_size']}, Patience: {cfg['train_patience']}")
 
         try:
             start_training(
@@ -416,21 +458,21 @@ def run_overnight(cfg):
                 batch_size=cfg["train_batch_size"],
                 patience=cfg["train_patience"],
             )
-            log("Training started, monitoring progress...")
+            log.info("Training started, monitoring progress...")
             success = wait_for_training(base_url)
             if success:
-                log("")
-                log("=" * 60)
-                log("  TRAINING COMPLETE — model updated!")
-                log("=" * 60)
+                log.info("")
+                log.info("=" * 60)
+                log.info("  TRAINING COMPLETE — model updated!")
+                log.info("=" * 60)
             else:
-                log("Training finished with errors. Check the Training tab for details.")
+                log.info("Training finished with errors. Check the Training tab for details.")
         except Exception as e:
-            log(f"ERROR starting training: {e}")
+            log.error(f"ERROR starting training: {e}")
             return False
 
-    log("")
-    log("Overnight run finished. You can close this window.")
+    log.info("")
+    log.info("Overnight run finished. You can close this window.")
     return True
 
 

pom.xml

@@ -67,6 +67,13 @@
             <version>3.45.1.0</version>
         </dependency>
 
+        <!-- WebSocket server for Unity/Godot client -->
+        <dependency>
+            <groupId>org.java-websocket</groupId>
+            <artifactId>Java-WebSocket</artifactId>
+            <version>1.5.6</version>
+        </dependency>
+
         <!-- Testing -->
         <dependency>
             <groupId>org.junit.jupiter</groupId>

pyproject.toml

@@ -10,8 +10,32 @@ readme = "README.md"
 requires-python = ">=3.10"
 license = {text = "MIT"}
 
+dependencies = [
+    "fastapi>=0.100.0",
+    "uvicorn[standard]>=0.23.0",
+    "pydantic>=2.0.0",
+    "httpx>=0.24.0",
+    "websockets>=11.0",
+]
+
+[project.optional-dependencies]
+dev = [
+    "pytest>=7.0",
+    "pytest-asyncio>=0.21",
+    "httpx>=0.24.0",
+    "ruff>=0.1.0",
+]
+
 [tool.setuptools.packages.find]
 where = ["src"]
 
 [project.scripts]
 commander-sim = "commander_ai_lab.lab.experiments:_cli_main"
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]
+pythonpath = [".", "src"]
+
+[tool.ruff]
+line-length = 120
+target-version = "py310"

pytest.ini

@@ -0,0 +1,24 @@
+[pytest]
+testpaths = tests
+python_files = test_*.py
+python_classes = Test* Bench*
+python_functions = test_*
+
+# Asyncio mode for pytest-asyncio
+asyncio_mode = auto
+
+# Markers
+markers =
+    slow: marks tests as slow (deselect with -m "not slow")
+    integration: marks integration tests
+    benchmark: marks benchmark tests
+    unit: marks fast unit tests
+
+# Output
+addopts =
+    --tb=short
+    -q
+
+# Coverage (used when running: pytest --cov)
+# pip install pytest-cov
+# pytest --cov=src/commander_ai_lab --cov-report=term-missing

requirements-web.txt

@@ -0,0 +1,11 @@
+# Commander AI Lab — Web server dependencies (Phase 5)
+fastapi>=0.110.0
+uvicorn[standard]>=0.29.0
+httpx>=0.27.0          # async Scryfall art proxy
+websockets>=12.0       # used by uvicorn WebSocket transport
+pydantic>=2.6.0
+
+# Dev / test
+pytest>=8.0
+pytest-asyncio>=0.23
+httpx>=0.27.0          # TestClient uses httpx transport

requirements.txt

@@ -0,0 +1,15 @@
+# Core runtime dependencies
+# Install with: pip install -r requirements.txt
+# Or for dev tools: pip install -e ".[dev]"
+
+fastapi>=0.100.0
+uvicorn[standard]>=0.23.0
+pydantic>=2.0.0
+httpx>=0.24.0
+websockets>=11.0
+
+# Deck Builder dependencies
+requests>=2.31.0
+beautifulsoup4>=4.12.0
+ollama>=0.2.0
+scrython>=2.0.0

routes/__init__.py

@@ -0,0 +1,6 @@
+"""
+routes package — API routers for Commander AI Lab.
+
+Each module defines a FastAPI APIRouter with real endpoint implementations.
+All shared helpers, models, and state live in routes.shared.
+"""

routes/coach.py

@@ -0,0 +1,671 @@
+"""
+routes/coach.py
+===============
+LLM-powered deck coaching endpoints:
+  GET  /api/coach/status
+  GET  /api/coach/decks
+  GET  /api/coach/decks/{deck_id}/report
+  POST /api/coach/decks/{deck_id}
+  GET  /api/coach/sessions
+  GET  /api/coach/sessions/{session_id}
+  POST /api/coach/embeddings/download
+  GET  /api/coach/embeddings/search
+  POST /api/coach/chat
+  POST /api/coach/apply
+  GET  /api/coach/cards-like
+  POST /api/coach/reports/generate
+
+Globals:
+  _coach_service, _coach_embeddings, _coach_llm
+  _deck_gen_v3, _deck_gen_v3_error
+  init_coach_service(), _build_deck_report_from_db()
+"""
+from __future__ import annotations
+
+import json
+import os
+import time
+import traceback
+from pathlib import Path
+from typing import Optional
+
+from fastapi import APIRouter, HTTPException
+from fastapi.responses import StreamingResponse
+from models.requests import CoachRequestBody, CoachChatMessage, CoachChatRequest, CoachApplyRequest, CoachGoalsRequest
+
+from routes.shared import (
+    CFG,
+    log_coach, log_deckgen,
+    _get_db_conn,
+    _compute_deck_analysis,
+)
+
+router = APIRouter(tags=["coach"])
+
+
+# ══════════════════════════════════════════════════════════════
+# Coach Service (LLM-powered deck coaching)
+# ══════════════════════════════════════════════════════════════
+
+# Global coach instances (initialized at startup)
+_coach_service = None
+_coach_embeddings = None
+_coach_llm = None
+_deck_gen_v3 = None  # V3 Deck Generator (Perplexity structured output)
+_deck_gen_v3_error = None  # Error message if V3 init failed
+
+
+def init_coach_service():
+    """Initialize the coach service with LLM client and embeddings."""
+    global _coach_service, _coach_embeddings, _coach_llm, _deck_gen_v3, _deck_gen_v3_error
+    try:
+        from coach.llm_client import LLMClient
+        from coach.embeddings import MTGEmbeddingIndex
+        from coach.coach_service import CoachService
+        from coach.config import ensure_dirs
+
+        ensure_dirs()
+
+        _coach_llm = LLMClient()
+        _coach_embeddings = MTGEmbeddingIndex()
+
+        # Try to load embeddings (non-blocking — download happens on first use)
+        try:
+            from coach.config import EMBEDDINGS_NPZ
+            if EMBEDDINGS_NPZ.exists():
+                _coach_embeddings.load()
+                log_coach.info(f"  Coach:        Embeddings loaded ({_coach_embeddings.card_count} cards)")
+            else:
+                log_coach.warning("  Coach:        Embeddings not yet downloaded (will download on first use)")
+        except Exception as e:
+            log_coach.error(f"  Coach:        Embeddings load failed: {e}")
+
+        _coach_service = CoachService(_coach_llm, _coach_embeddings)
+
+        # Check LLM connection
+        llm_status = _coach_llm.check_connection()
+        if llm_status.get("connected"):
+            log_coach.info(f"  Coach LLM:    Connected ({llm_status.get('active_model', 'unknown')})")
+        else:
+            log_coach.warning(f"  Coach LLM:    Not connected (start Ollama on 192.168.0.240:11434)")
+
+        # Initialize V3 Deck Generator (Ollama-based)
+        _deck_gen_v3_error = None
+        try:
+            from src.commander_ai_lab.deck_builder.adapter import DeckBuilderAdapter
+            _deck_gen_v3 = DeckBuilderAdapter(
+                db_conn_factory=_get_db_conn,
+                model="gpt-oss:20b",
+            )
+            log_deckgen.info("  Deck Gen V3:  Initialized (Ollama DeckBuilderAdapter)")
+        except ImportError as e:
+            _deck_gen_v3_error = f"Missing dependency: {e}"
+            log_deckgen.error(f"  Deck Gen V3:  {_deck_gen_v3_error}")
+            _deck_gen_v3 = None
+        except Exception as e:
+            _deck_gen_v3_error = str(e)
+            log_deckgen.error(f"  Deck Gen V3:  Failed to initialize: {e}")
+            _deck_gen_v3 = None
+
+    except Exception as e:
+        log_coach.error(f"  Coach:        Failed to initialize: {e}\n{traceback.format_exc()}")
+        _coach_service = None
+
+
+@router.get("/api/coach/status")
+async def coach_status():
+    """Check coach subsystem health."""
+    if _coach_service is None:
+        return {"llmConnected": False, "embeddingsLoaded": False,
+                "embeddingCards": 0, "deckReportsAvailable": 0,
+                "error": "Coach service not initialized"}
+    status = _coach_service.get_status()
+    return status.model_dump()
+
+
+@router.get("/api/coach/decks")
+async def coach_list_decks():
+    """List all decks available for coaching (from deck builder DB)."""
+    conn = _get_db_conn()
+    rows = conn.execute(
+        """SELECT d.id as deck_id, d.name as deck_name, d.commander_name,
+                  COUNT(dc.id) as card_count
+           FROM decks d
+           LEFT JOIN deck_cards dc ON dc.deck_id = d.id
+           GROUP BY d.id
+           ORDER BY d.name"""
+    ).fetchall()
+
+    # Also get report availability from coach service
+    report_ids = set()
+    if _coach_service:
+        report_ids = set(_coach_service.list_deck_reports())
+
+    decks = []
+    for r in rows:
+        deck_name = r["deck_name"]
+        # Check if a report exists (by deck name slug match)
+        has_report = any(
+            deck_name.lower().replace(" ", "-") == rid.lower() or
+            deck_name.lower() == rid.lower()
+            for rid in report_ids
+        )
+        decks.append({
+            "deck_id": r["deck_id"],
+            "deck_name": r["deck_name"],
+            "commander": r["commander_name"] or "",
+            "card_count": r["card_count"],
+            "has_report": has_report,
+            "report_count": 1 if has_report else 0,
+            "last_report_date": None,
+        })
+    return decks
+
+
+@router.get("/api/coach/decks/{deck_id}/report")
+async def coach_get_report(deck_id: str):
+    """Get the latest DeckReport for a deck."""
+    if _coach_service is None:
+        raise HTTPException(500, "Coach service not initialized")
+    report = _coach_service.load_deck_report(deck_id)
+    if report is None:
+        raise HTTPException(404, f"Deck report not found: {deck_id}")
+    return report.model_dump()
+
+
+@router.post("/api/coach/decks/{deck_id}")
+async def coach_run_session(deck_id: str, body: CoachRequestBody = None):
+    """Trigger a coaching session for a deck."""
+    if _coach_service is None:
+        raise HTTPException(500, "Coach service not initialized")
+
+    # Load embeddings on first use if not loaded
+    if _coach_embeddings and not _coach_embeddings.loaded:
+        try:
+            _coach_embeddings.load(force_download=True)
+        except Exception as e:
+            log_coach.error(f"  Coach: Embeddings download failed: {e}")
+
+    goals = None
+    if body and body.goals:
+        from coach.models import CoachGoals
+        try:
+            goals = CoachGoals(**body.goals)
+        except Exception:
+            goals = None
+
+    # Build a fallback DeckReport from the DB if no simulation report exists
+    fallback_report = None
+    try:
+        fallback_report = _build_deck_report_from_db(deck_id)
+    except Exception as e:
+        log_coach.error(f"  Coach: Fallback report build failed for '{deck_id}': {e}")
+
+    try:
+        session = await _coach_service.run_coaching_session(deck_id, goals, fallback_report=fallback_report)
+        return session.model_dump()
+    except ValueError as e:
+        raise HTTPException(404, str(e))
+    except ConnectionError as e:
+        raise HTTPException(503, f"LLM connection failed: {e}")
+    except Exception as e:
+        raise HTTPException(500, f"Coach session failed: {e}")
+
+
+def _build_deck_report_from_db(deck_slug: str):
+    """
+    Build a lightweight DeckReport from the deck builder DB when no
+    simulation report exists. Allows the coach to analyze deck composition
+    even without simulation data.
+    """
+    from coach.models import DeckReport, CardPerformance, DeckStructure
+    conn = _get_db_conn()
+
+    # --- Input validation ---
+    import re
+    if not deck_slug or len(deck_slug) > 200:
+        return None
+    deck_slug = re.sub(r'[^a-zA-Z0-9 _-]', '', deck_slug)
+    if not deck_slug:
+        return None
+
+    # Find the deck by parameterized query instead of full-table scan
+    slug_lower = deck_slug.lower()
+    slug_hyphenated = slug_lower.replace(' ', '-')
+    clean_slug = re.sub(r'[^a-z0-9]+', '-', slug_lower).strip('-')
+
+    if slug_lower.isdigit():
+        # Numeric slug -- match by ID directly
+        matched_deck = conn.execute(
+            "SELECT id, name, commander_name, color_identity FROM decks WHERE id = ?",
+            (int(slug_lower),)
+        ).fetchone()
+    else:
+        # Try exact name match, hyphenated match, or clean-slug match
+        matched_deck = conn.execute(
+            """SELECT id, name, commander_name, color_identity FROM decks
+               WHERE LOWER(name) = ?
+                  OR LOWER(REPLACE(name, ' ', '-')) = ?
+                  OR LOWER(REPLACE(name, ' ', '-')) = ?
+               LIMIT 1""",
+            (slug_lower, slug_hyphenated, clean_slug)
+        ).fetchone()
+
+    if matched_deck is None:
+        return None
+
+    deck_id = matched_deck["id"]
+    deck_name = matched_deck["name"]
+    commander = matched_deck["commander_name"] or ""
+
+    # Parse color identity
+    ci_raw = matched_deck["color_identity"] or "[]"
+    try:
+        color_identity = json.loads(ci_raw) if isinstance(ci_raw, str) else ci_raw
+    except Exception:
+        color_identity = []
+
+    # Load all cards in this deck
+    card_rows = conn.execute(
+        """SELECT dc.card_name, dc.quantity, dc.is_commander, dc.role_tag,
+                  ce.type_line, ce.cmc, ce.oracle_text
+           FROM deck_cards dc
+           LEFT JOIN collection_entries ce ON LOWER(dc.card_name) = LOWER(ce.name)
+           WHERE dc.deck_id = ?""",
+        (deck_id,)
+    ).fetchall()
+
+    cards = []
+    type_counts = {}
+    cmc_buckets = [0] * 8
+    land_count = 0
+
+    for cr in card_rows:
+        card_name = cr["card_name"] or ""
+        type_line = cr["type_line"] or ""
+        cmc = cr["cmc"] or 0
+        qty = cr["quantity"] or 1
+        role_tag = cr["role_tag"] or ""
+
+        # Build tags from type_line and role_tag
+        tags = []
+        if role_tag:
+            tags.append(role_tag)
+        if "Land" in type_line:
+            tags.append("land")
+            land_count += qty
+        elif "Creature" in type_line:
+            tags.append("creature")
+        elif "Instant" in type_line:
+            tags.append("instant")
+        elif "Sorcery" in type_line:
+            tags.append("sorcery")
+        elif "Artifact" in type_line:
+            tags.append("artifact")
+        elif "Enchantment" in type_line:
+            tags.append("enchantment")
+        elif "Planeswalker" in type_line:
+            tags.append("planeswalker")
+
+        # Type count
+        for t in ["Creature", "Instant", "Sorcery", "Artifact", "Enchantment", "Planeswalker", "Land"]:
+            if t in type_line:
+                type_counts[t] = type_counts.get(t, 0) + qty
+
+        # CMC bucket
+        bucket = min(int(cmc), 7)
+        cmc_buckets[bucket] += qty
+
+        # CardPerformance with zeroed-out sim stats
+        cards.append(CardPerformance(
+            name=card_name,
+            drawnRate=0.0,
+            castRate=0.0,
+            impactScore=0.0,
+            tags=tags,
+        ))
+
+    slug = deck_slug
+    return DeckReport(
+        deckId=slug,
+        commander=commander,
+        colorIdentity=color_identity,
+        cards=cards,
+        structure=DeckStructure(
+            landCount=land_count,
+            curveBuckets=cmc_buckets,
+            cardTypeCounts=type_counts,
+        ),
+    )
+
+
+@router.get("/api/coach/sessions")
+async def coach_list_sessions(deck_id: str = None):
+    """List all coaching sessions, optionally filtered by deck."""
+    if _coach_service is None:
+        raise HTTPException(500, "Coach service not initialized")
+    return {"sessions": _coach_service.list_sessions(deck_id)}
+
+
+@router.get("/api/coach/sessions/{session_id}")
+async def coach_get_session(session_id: str):
+    """Get a specific coaching session."""
+    if _coach_service is None:
+        raise HTTPException(500, "Coach service not initialized")
+    session = _coach_service.load_session(session_id)
+    if session is None:
+        raise HTTPException(404, f"Session not found: {session_id}")
+    return session.model_dump()
+
+
+@router.post("/api/coach/embeddings/download")
+async def coach_download_embeddings():
+    """Download and convert MTG card embeddings from HuggingFace."""
+    if _coach_embeddings is None:
+        raise HTTPException(500, "Coach service not initialized")
+    try:
+        _coach_embeddings.load(force_download=True)
+        return {
+            "success": True,
+            "cards": _coach_embeddings.card_count,
+            "message": f"Loaded {_coach_embeddings.card_count} card embeddings"
+        }
+    except Exception as e:
+        raise HTTPException(500, f"Failed to download embeddings: {e}")
+
+
+@router.get("/api/coach/embeddings/search")
+async def coach_search_similar(card: str, colors: str = None, top_n: int = 10):
+    """Search for similar cards by name."""
+    if _coach_embeddings is None or not _coach_embeddings.loaded:
+        raise HTTPException(503, "Embeddings not loaded")
+    color_filter = list(colors.upper()) if colors else None
+    matches = _coach_embeddings.search_similar(
+        query_card=card, color_filter=color_filter, top_n=top_n
+    )
+    return {"query": card, "matches": [m.to_dict() for m in matches]}
+
+
+@router.post("/api/coach/chat")
+async def coach_chat(body: CoachChatRequest):
+    """Multi-turn coaching chat. Sends conversation history to LLM and returns response."""
+    if _coach_service is None:
+        raise HTTPException(500, "Coach service not initialized")
+
+    # Load deck report for context
+    report = _coach_service.load_deck_report(body.deck_id)
+
+    # Build system prompt with report context
+    from coach.prompt_template import build_system_prompt
+    from coach.models import CoachGoals
+
+    goals = None
+    if body.goals:
+        try:
+            goals = CoachGoals(**body.goals)
+        except Exception:
+            goals = None
+
+    system_prompt = ""
+    if report:
+        system_prompt = build_system_prompt(report, goals)
+    else:
+        system_prompt = (
+            "You are an expert Magic: The Gathering Commander deck coach. "
+            "Answer questions about deck building, strategy, card choices, and game theory. "
+            "Be specific and actionable in your advice."
+        )
+
+    # Build messages array for LLM
+    messages = [{"role": "system", "content": system_prompt}]
+    for msg in body.messages:
+        messages.append({"role": msg.get("role", "user"), "content": msg.get("content", "")})
+
+    # Call LLM with multi-turn messages
+    try:
+        import json
+        from urllib.request import urlopen, Request as UrlRequest
+        from coach.config import LLM_URL, LLM_TIMEOUT
+
+        model_name = _coach_llm._resolve_model()
+        llm_body = {
+            "model": model_name,
+            "messages": messages,
+            "temperature": 0.7,
+            "max_tokens": 4096,
+        }
+
+        if body.stream:
+            # SSE streaming response
+            llm_body["stream"] = True
+
+            async def generate():
+                import asyncio
+                def _stream():
+                    req = UrlRequest(
+                        f"{_coach_llm.base_url}/chat/completions",
+                        data=json.dumps(llm_body).encode("utf-8"),
+                        headers={"Content-Type": "application/json"},
+                        method="POST",
+                    )
+                    import http.client
+                    import urllib.parse
+                    parsed = urllib.parse.urlparse(f"{_coach_llm.base_url}/chat/completions")
+                    conn = http.client.HTTPConnection(parsed.hostname, parsed.port, timeout=LLM_TIMEOUT)
+                    conn.request("POST", parsed.path, body=json.dumps(llm_body).encode("utf-8"),
+                                 headers={"Content-Type": "application/json"})
+                    resp = conn.getresponse()
+                    chunks = []
+                    while True:
+                        line = resp.readline()
+                        if not line:
+                            break
+                        line = line.decode("utf-8").strip()
+                        if line.startswith("data: "):
+                            data = line[6:]
+                            if data == "[DONE]":
+                                chunks.append("[DONE]")
+                                break
+                            chunks.append(data)
+                    conn.close()
+                    return chunks
+
+                loop = asyncio.get_event_loop()
+                chunks = await loop.run_in_executor(None, _stream)
+
+                for chunk_str in chunks:
+                    if chunk_str == "[DONE]":
+                        yield f"data: [DONE]\n\n"
+                        break
+                    try:
+                        chunk = json.loads(chunk_str)
+                        delta = chunk.get("choices", [{}])[0].get("delta", {})
+                        content = delta.get("content", "")
+                        if content:
+                            # Strip think tags from streaming chunks
+                            yield f"data: {json.dumps({'content': content})}\n\n"
+                    except json.JSONDecodeError:
+                        continue
+
+            return StreamingResponse(generate(), media_type="text/event-stream")
+        else:
+            # Non-streaming: regular call
+            req = UrlRequest(
+                f"{_coach_llm.base_url}/chat/completions",
+                data=json.dumps(llm_body).encode("utf-8"),
+                headers={"Content-Type": "application/json"},
+                method="POST",
+            )
+
+            import asyncio
+            loop = asyncio.get_event_loop()
+            def _call():
+                with urlopen(req, timeout=LLM_TIMEOUT) as resp:
+                    return json.loads(resp.read().decode("utf-8"))
+
+            raw = await loop.run_in_executor(None, _call)
+
+            content = raw.get("choices", [{}])[0].get("message", {}).get("content", "")
+            # Strip think tags
+            content = _coach_llm._strip_think_tags(content)
+            usage = raw.get("usage", {})
+
+            return {
+                "content": content,
+                "model": raw.get("model", ""),
+                "prompt_tokens": usage.get("prompt_tokens", 0),
+                "completion_tokens": usage.get("completion_tokens", 0),
+            }
+
+    except Exception as e:
+        raise HTTPException(500, f"Chat failed: {e}")
+
+
+@router.post("/api/coach/apply")
+async def coach_apply_suggestions(body: CoachApplyRequest):
+    """Apply accepted coaching suggestions to a deck — remove cuts, add adds."""
+    conn = _get_db_conn()
+
+    # Verify deck exists
+    deck = conn.execute("SELECT * FROM decks WHERE id = ?", (body.deck_id,)).fetchone()
+    if not deck:
+        raise HTTPException(404, f"Deck {body.deck_id} not found")
+
+    results = {"cuts": [], "adds": [], "errors": []}
+
+    # Process cuts: remove cards by name
+    for card_name in body.accepted_cuts:
+        row = conn.execute(
+            "SELECT id FROM deck_cards WHERE deck_id = ? AND card_name = ? LIMIT 1",
+            (body.deck_id, card_name)
+        ).fetchone()
+        if row:
+            conn.execute("DELETE FROM deck_cards WHERE id = ?", (row["id"],))
+            results["cuts"].append({"name": card_name, "status": "removed"})
+        else:
+            # Try case-insensitive
+            row = conn.execute(
+                "SELECT id, card_name FROM deck_cards WHERE deck_id = ? AND LOWER(card_name) = LOWER(?) LIMIT 1",
+                (body.deck_id, card_name)
+            ).fetchone()
+            if row:
+                conn.execute("DELETE FROM deck_cards WHERE id = ?", (row["id"],))
+                results["cuts"].append({"name": row["card_name"], "status": "removed"})
+            else:
+                results["errors"].append({"name": card_name, "error": "Card not found in deck"})
+
+    # Process adds: look up in collection first, then Scryfall
+    for card_name in body.accepted_adds:
+        # Try to find in collection
+        ce = conn.execute(
+            "SELECT scryfall_id, name FROM collection_entries WHERE LOWER(name) = LOWER(?) LIMIT 1",
+            (card_name,)
+        ).fetchone()
+
+        scryfall_id = None
+        resolved_name = card_name
+
+        if ce:
+            scryfall_id = ce["scryfall_id"]
+            resolved_name = ce["name"]
+        else:
+            # Try Scryfall API lookup
+            try:
+                import urllib.parse
+                from urllib.request import urlopen, Request as UrlRequest
+                encoded = urllib.parse.quote(card_name)
+                scry_req = UrlRequest(
+                    f"https://api.scryfall.com/cards/named?fuzzy={encoded}",
+                    headers={"User-Agent": "commander-ai-lab/1.0"}
+                )
+                with urlopen(scry_req, timeout=10) as resp:
+                    card_data = json.loads(resp.read().decode("utf-8"))
+                    scryfall_id = card_data.get("id")
+                    resolved_name = card_data.get("name", card_name)
+            except Exception:
+                pass
+
+        if scryfall_id:
+            # Check if already in deck
+            existing = conn.execute(
+                "SELECT id FROM deck_cards WHERE deck_id = ? AND scryfall_id = ?",
+                (body.deck_id, scryfall_id)
+            ).fetchone()
+            if not existing:
+                conn.execute(
+                    "INSERT INTO deck_cards (deck_id, scryfall_id, card_name, quantity) VALUES (?, ?, ?, 1)",
+                    (body.deck_id, scryfall_id, resolved_name)
+                )
+                results["adds"].append({"name": resolved_name, "scryfall_id": scryfall_id, "status": "added"})
+            else:
+                results["adds"].append({"name": resolved_name, "status": "already_in_deck"})
+        else:
+            results["errors"].append({"name": card_name, "error": "Could not resolve card"})
+
+    conn.execute("UPDATE decks SET updated_at = datetime('now') WHERE id = ?", (body.deck_id,))
+    conn.commit()
+
+    results["total_cuts"] = len(results["cuts"])
+    results["total_adds"] = len([a for a in results["adds"] if a["status"] == "added"])
+    return results
+
+
+@router.get("/api/coach/cards-like")
+async def coach_cards_like(card: str, colors: str = None, top_n: int = 10):
+    """Find cards similar to a given card using embeddings. UI-friendly version."""
+    if _coach_embeddings is None or not _coach_embeddings.loaded:
+        raise HTTPException(503, "Embeddings not loaded")
+
+    color_filter = list(colors.upper()) if colors else None
+    matches = _coach_embeddings.search_similar(
+        query_card=card, color_filter=color_filter, top_n=top_n
+    )
+
+    conn = _get_db_conn()
+    results = []
+    for m in matches:
+        # Check if card is in collection
+        owned = conn.execute(
+            "SELECT SUM(quantity) as qty FROM collection_entries WHERE LOWER(name) = LOWER(?)",
+            (m.name,)
+        ).fetchone()
+        owned_qty = (owned["qty"] or 0) if owned else 0
+
+        # Get image URL and price
+        ce = conn.execute(
+            "SELECT image_url, tcg_price FROM collection_entries WHERE LOWER(name) = LOWER(?) LIMIT 1",
+            (m.name,)
+        ).fetchone()
+
+        results.append({
+            "name": m.name,
+            "similarity": round(m.similarity, 4),
+            "types": m.types,
+            "mana_value": m.mana_value,
+            "mana_cost": m.mana_cost,
+            "text": m.text[:200] if m.text else "",
+            "owned_qty": owned_qty,
+            "image_url": ce["image_url"] if ce else None,
+            "tcg_price": ce["tcg_price"] if ce else None,
+        })
+
+    return {"query": card, "results": results}
+
+
+@router.post("/api/coach/reports/generate")
+async def coach_generate_reports():
+    """Rebuild all deck reports from batch result JSONs in results/."""
+    try:
+        from coach.report_generator import generate_deck_reports
+        lab_root = Path(__file__).parent
+        results_dir = str(lab_root / CFG.results_dir)
+        reports_dir = str(lab_root / "deck-reports")
+        updated = generate_deck_reports(results_dir, reports_dir)
+        return {
+            "status": "ok",
+            "decksUpdated": updated,
+            "count": len(updated),
+            "message": f"Generated reports for {len(updated)} decks" if updated else "No batch results found",
+        }
+    except Exception as e:
+        raise HTTPException(500, f"Report generation failed: {e}")

routes/collection.py

@@ -0,0 +1,487 @@
+"""
+routes/collection.py
+====================
+Collection management endpoints:
+  GET    /api/collection
+  GET    /api/collection/export
+  GET    /api/collection/sets
+  GET    /api/collection/keywords
+  POST   /api/collection/import
+  GET    /api/collection/{cardId}
+  PATCH  /api/collection/{cardId}
+  GET    /api/cache/scryfall
+  DELETE /api/cache/scryfall
+  POST   /api/cache/scryfall/evict
+"""
+from __future__ import annotations
+
+import csv
+import io
+import json
+import re
+from datetime import datetime
+from typing import Optional
+
+from fastapi import APIRouter, HTTPException
+from fastapi.responses import JSONResponse, StreamingResponse
+
+from routes.shared import (
+    CFG,
+    _get_db_conn,
+    _row_to_dict,
+    _add_image_url,
+    _build_collection_filters,
+    _detect_card_roles,
+    _classify_card_type,
+    _enrich_from_scryfall,
+    _fetch_scryfall_api,
+    _parse_finish,
+    _parse_text_line,
+    _parse_csv_content,
+    _auto_infer_mapping,
+    _scryfall_cache,
+    _scryfall_rate_limit,
+    _snake_to_camel,
+    VALID_SORT_FIELDS,
+    _JSON_FIELDS,
+    SCRYFALL_CACHE_DB_PATH,
+    SCRYFALL_CACHE_TTL_SECONDS,
+    _API_HEADERS,
+    log_collect,
+    log_cache,
+)
+
+router = APIRouter(tags=["collection"])
+
+
+@router.get("/api/collection/export")
+async def export_collection(
+    format: str = "INTERNAL_CSV",
+    q: Optional[str] = None,
+    colors: Optional[str] = None,
+    types: Optional[str] = None,
+    isLegendary: Optional[bool] = None,
+    isBasic: Optional[bool] = None,
+    isGameChanger: Optional[bool] = None,
+    highSalt: Optional[bool] = None,
+    finish: Optional[str] = None,
+    cmcMin: Optional[float] = None,
+    cmcMax: Optional[float] = None,
+    priceMin: Optional[float] = None,
+    priceMax: Optional[float] = None,
+    category: Optional[str] = None,
+):
+    """Export collection in various formats with optional filters."""
+    where_str, params = _build_collection_filters(
+        q=q, colors=colors, types=types, isLegendary=isLegendary,
+        isBasic=isBasic, isGameChanger=isGameChanger, highSalt=highSalt,
+        finish=finish, cmcMin=cmcMin, cmcMax=cmcMax,
+        priceMin=priceMin, priceMax=priceMax, category=category,
+    )
+
+    sql = f"SELECT * FROM collection_entries {where_str} ORDER BY name ASC"
+    conn = _get_db_conn()
+    rows = [_row_to_dict(r) for r in conn.execute(sql, params).fetchall()]
+
+    fmt = format.upper()
+    output = io.StringIO()
+
+    if fmt == "INTERNAL_CSV":
+        if rows:
+            fieldnames = list(rows[0].keys())
+        else:
+            fieldnames = [
+                "id", "name", "type_line", "subtypes", "is_legendary", "is_basic",
+                "color_identity", "cmc", "oracle_text", "keywords", "tcg_price",
+                "salt_score", "is_game_changer", "category", "scryfall_id", "tcgplayer_id",
+                "quantity", "finish", "condition", "language", "notes", "tags",
+                "set_code", "collector_number", "created_at", "updated_at",
+            ]
+        writer = csv.DictWriter(output, fieldnames=fieldnames)
+        writer.writeheader()
+        writer.writerows(rows)
+        filename = "collection_export.csv"
+        media_type = "text/csv"
+
+    elif fmt == "MOXFIELD_CSV":
+        fieldnames = ["Count", "Name", "Edition", "Condition", "Language", "Foil", "Collector Number"]
+        writer = csv.DictWriter(output, fieldnames=fieldnames)
+        writer.writeheader()
+        for row in rows:
+            finish_val = row.get("finish", "NORMAL").upper()
+            foil_val = "foil" if finish_val == "FOIL" else ("etched" if finish_val == "ETCHED" else "")
+            writer.writerow({
+                "Count": row.get("quantity", 1),
+                "Name": row.get("name", ""),
+                "Edition": row.get("set_code", "").upper(),
+                "Condition": row.get("condition", ""),
+                "Language": row.get("language", ""),
+                "Foil": foil_val,
+                "Collector Number": row.get("collector_number", ""),
+            })
+        filename = "collection_moxfield.csv"
+        media_type = "text/csv"
+
+    elif fmt == "TEXT":
+        lines = []
+        for row in rows:
+            qty = row.get("quantity", 1)
+            name = row.get("name", "")
+            set_code = row.get("set_code", "").upper()
+            coll_num = row.get("collector_number", "")
+            if set_code and coll_num:
+                lines.append(f"{qty} {name} ({set_code}) {coll_num}")
+            elif set_code:
+                lines.append(f"{qty} {name} ({set_code})")
+            else:
+                lines.append(f"{qty} {name}")
+        output.write("\n".join(lines))
+        filename = "collection_export.txt"
+        media_type = "text/plain"
+
+    else:
+        raise HTTPException(400, f"Unknown format: {format}. Use INTERNAL_CSV, MOXFIELD_CSV, or TEXT.")
+
+    content = output.getvalue()
+
+    def iter_content():
+        yield content.encode("utf-8")
+
+    return StreamingResponse(
+        iter_content(),
+        media_type=media_type,
+        headers={"Content-Disposition": f'attachment; filename="{filename}"'},
+    )
+
+
+@router.get("/api/collection/sets")
+async def collection_sets():
+    """Return distinct set codes + names in the collection (for filter autocomplete)."""
+    conn = _get_db_conn()
+    rows = conn.execute(
+        "SELECT DISTINCT UPPER(set_code) as code, set_name as name FROM collection_entries WHERE set_code != '' ORDER BY set_name ASC"
+    ).fetchall()
+    result = []
+    for r in rows:
+        code = r["code"] if "code" in r.keys() else r[0]
+        name = r["name"] if "name" in r.keys() else r[1]
+        if code:
+            result.append({"code": code, "name": name or code})
+    return JSONResponse(result)
+
+
+@router.get("/api/collection/keywords")
+async def collection_keywords():
+    """Return distinct keywords found in the collection."""
+    conn = _get_db_conn()
+    rows = conn.execute("SELECT keywords FROM collection_entries WHERE keywords != '' AND keywords != '[]'").fetchall()
+    kw_set = set()
+    for r in rows:
+        try:
+            kws = json.loads(r["keywords"]) if isinstance(r["keywords"], str) else r["keywords"]
+            if isinstance(kws, list):
+                for k in kws:
+                    if k:
+                        kw_set.add(k)
+        except (json.JSONDecodeError, TypeError):
+            pass
+    return JSONResponse(sorted(kw_set))
+
+
+@router.get("/api/collection")
+async def list_collection(
+    page: int = 1,
+    pageSize: int = 50,
+    q: Optional[str] = None,
+    sortField: Optional[str] = None,
+    sortDir: Optional[str] = None,
+    colors: Optional[str] = None,
+    types: Optional[str] = None,
+    isLegendary: Optional[bool] = None,
+    isBasic: Optional[bool] = None,
+    isGameChanger: Optional[bool] = None,
+    highSalt: Optional[bool] = None,
+    finish: Optional[str] = None,
+    cmcMin: Optional[float] = None,
+    cmcMax: Optional[float] = None,
+    priceMin: Optional[float] = None,
+    priceMax: Optional[float] = None,
+    category: Optional[str] = None,
+    deck_id: Optional[int] = None,
+    rarity: Optional[str] = None,
+    setCode: Optional[str] = None,
+    powerMin: Optional[str] = None,
+    powerMax: Optional[str] = None,
+    toughMin: Optional[str] = None,
+    toughMax: Optional[str] = None,
+    keyword: Optional[str] = None,
+    edhrecMin: Optional[int] = None,
+    edhrecMax: Optional[int] = None,
+    qtyMin: Optional[int] = None,
+    qtyMax: Optional[int] = None,
+):
+    """List collection entries with search, sort, and filter.
+    Optional deck_id: when provided, includes in_deck_quantity for each result.
+    """
+    where_str, params = _build_collection_filters(
+        q=q, colors=colors, types=types, isLegendary=isLegendary,
+        isBasic=isBasic, isGameChanger=isGameChanger, highSalt=highSalt,
+        finish=finish, cmcMin=cmcMin, cmcMax=cmcMax,
+        priceMin=priceMin, priceMax=priceMax, category=category,
+        rarity=rarity, setCode=setCode, powerMin=powerMin, powerMax=powerMax,
+        toughMin=toughMin, toughMax=toughMax, keyword=keyword,
+        edhrecMin=edhrecMin, edhrecMax=edhrecMax, qtyMin=qtyMin, qtyMax=qtyMax,
+    )
+
+    # Validate sort
+    sort_field = sortField if sortField in VALID_SORT_FIELDS else "name"
+    sort_dir = "DESC" if (sortDir or "").upper() == "DESC" else "ASC"
+
+    conn = _get_db_conn()
+
+    # Count total
+    count_sql = f"SELECT COUNT(*) FROM collection_entries {where_str}"
+    total = conn.execute(count_sql, params).fetchone()[0]
+
+    # Paginate
+    offset = (max(1, page) - 1) * pageSize
+    data_sql = f"""
+        SELECT * FROM collection_entries {where_str}
+        ORDER BY {sort_field} {sort_dir}
+        LIMIT ? OFFSET ?
+    """
+    rows = conn.execute(data_sql, params + [pageSize, offset]).fetchall()
+    items = [_add_image_url(_row_to_dict(r)) for r in rows]
+
+    # If deck_id provided, annotate each item with how many copies are already in that deck
+    if deck_id is not None:
+        # Build a map of scryfall_id -> quantity in deck
+        deck_card_rows = conn.execute(
+            "SELECT scryfall_id, SUM(quantity) as qty FROM deck_cards WHERE deck_id = ? GROUP BY scryfall_id",
+            (deck_id,)
+        ).fetchall()
+        in_deck_map = {row["scryfall_id"]: row["qty"] for row in deck_card_rows}
+        for item in items:
+            item["in_deck_quantity"] = in_deck_map.get(item.get("scryfall_id", ""), 0)
+
+    return {"items": items, "page": page, "pageSize": pageSize, "total": total}
+
+
+@router.get("/api/collection/{cardId}")
+async def get_collection_card(cardId: int):
+    """Get a single collection entry by ID."""
+    conn = _get_db_conn()
+    row = conn.execute("SELECT * FROM collection_entries WHERE id = ?", (cardId,)).fetchone()
+    if not row:
+        raise HTTPException(404, f"Card with id {cardId} not found")
+    return _add_image_url(_row_to_dict(row))
+
+
+@router.patch("/api/collection/{cardId}")
+async def update_collection_card(cardId: int, body: dict):
+    """Update mutable fields of a collection entry."""
+    conn = _get_db_conn()
+    row = conn.execute("SELECT * FROM collection_entries WHERE id = ?", (cardId,)).fetchone()
+    if not row:
+        raise HTTPException(404, f"Card with id {cardId} not found")
+
+    updates = {}
+    if "category" in body:
+        cat = body["category"]
+        if isinstance(cat, list):
+            updates["category"] = json.dumps(cat)
+        else:
+            updates["category"] = cat
+    if "tags" in body:
+        updates["tags"] = str(body["tags"])
+    if "notes" in body:
+        updates["notes"] = str(body["notes"])
+    if "finish" in body:
+        finish_val = str(body["finish"]).upper()
+        if finish_val not in ("NORMAL", "FOIL", "ETCHED"):
+            raise HTTPException(400, "finish must be NORMAL, FOIL, or ETCHED")
+        updates["finish"] = finish_val
+
+    if not updates:
+        return _add_image_url(_row_to_dict(row))
+
+    updates["updated_at"] = datetime.utcnow().strftime("%Y-%m-%d %H:%M:%S")
+    set_clause = ", ".join(f"{k} = ?" for k in updates)
+    values = list(updates.values()) + [cardId]
+    conn.execute(f"UPDATE collection_entries SET {set_clause} WHERE id = ?", values)
+    conn.commit()
+
+    updated_row = conn.execute("SELECT * FROM collection_entries WHERE id = ?", (cardId,)).fetchone()
+    return _add_image_url(_row_to_dict(updated_row))
+
+
+# ══════════════════════════════════════════════════════════════
+# Collection Import Endpoint
+# ══════════════════════════════════════════════════════════════
+
+@router.post("/api/collection/import")
+async def import_collection(body: dict):
+    """
+    Import cards into the collection from CSV, Moxfield, Archidekt, or plain text.
+
+    Body:
+      {
+        "source": "CSV" | "MOXFIELD" | "ARCHIDEKT" | "TEXT",
+        "mode": "MERGE" | "REPLACE",
+        "content": "...csv or text content...",
+        "mapping": { ... optional column mapping ... }
+      }
+    """
+    source = str(body.get("source", "CSV")).upper()
+    mode = str(body.get("mode", "MERGE")).upper()
+    content = str(body.get("content", ""))
+    mapping = body.get("mapping", {})
+
+    if not content.strip():
+        raise HTTPException(400, "content is required")
+
+    imported_count = 0
+    updated_count = 0
+    failed_count = 0
+    errors = []
+
+    conn = _get_db_conn()
+
+    # REPLACE mode: clear existing entries
+    if mode == "REPLACE":
+        conn.execute("DELETE FROM collection_entries")
+        conn.commit()
+
+    # Parse the content into raw rows
+    try:
+        raw_rows = _parse_csv_content(content, source, mapping if mapping else None)
+    except Exception as e:
+        raise HTTPException(400, f"Failed to parse content: {str(e)}")
+
+    for raw in raw_rows:
+        name = raw.get("name", "").strip()
+        if not name:
+            continue
+
+        quantity = int(raw.get("quantity", 1))
+        set_code = raw.get("set_code", "").lower()
+        collector_number = raw.get("collector_number", "")
+        finish = _parse_finish(raw.get("finish", ""))
+        condition = raw.get("condition", "")
+        language = raw.get("language", "")
+        notes = raw.get("notes", "")
+        tags = raw.get("tags", "")
+
+        # Scryfall enrichment
+        try:
+            enriched = _enrich_from_scryfall(name, set_code, collector_number)
+        except Exception as e:
+            errors.append(f"Scryfall error for '{name}': {str(e)}")
+            failed_count += 1
+            continue
+
+        if not enriched or "_error" in enriched:
+            err_msg = enriched.get("_error", f"Card not found on Scryfall: '{name}'") if enriched else f"Card not found on Scryfall: '{name}'"
+            errors.append(err_msg)
+            failed_count += 1
+            continue
+
+        # Use Scryfall-resolved name if available
+        resolved_name = enriched.get("name", name)
+
+        # Check for existing entry (identity key)
+        existing = conn.execute(
+            """SELECT id, quantity FROM collection_entries
+               WHERE name = ? AND set_code = ? AND collector_number = ? AND finish = ?""",
+            (resolved_name, set_code, collector_number, finish),
+        ).fetchone()
+
+        if existing and mode == "MERGE":
+            # Update quantity
+            new_qty = existing["quantity"] + quantity
+            conn.execute(
+                "UPDATE collection_entries SET quantity = ?, updated_at = datetime('now') WHERE id = ?",
+                (new_qty, existing["id"]),
+            )
+            conn.commit()
+            updated_count += 1
+        else:
+            # Insert new row
+            conn.execute(
+                """INSERT INTO collection_entries
+                   (name, type_line, subtypes, is_legendary, is_basic, color_identity,
+                    cmc, mana_cost, oracle_text, keywords,
+                    power, toughness, rarity, set_name, edhrec_rank,
+                    tcg_price, salt_score, is_game_changer,
+                    category, scryfall_id, tcgplayer_id,
+                    quantity, finish, condition, language, notes, tags,
+                    set_code, collector_number)
+                   VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)""",
+                (
+                    resolved_name,
+                    enriched.get("type_line", ""),
+                    enriched.get("subtypes", "[]"),
+                    enriched.get("is_legendary", 0),
+                    enriched.get("is_basic", 0),
+                    enriched.get("color_identity", "[]"),
+                    enriched.get("cmc", 0.0),
+                    enriched.get("mana_cost", ""),
+                    enriched.get("oracle_text", ""),
+                    enriched.get("keywords", "[]"),
+                    enriched.get("power", ""),
+                    enriched.get("toughness", ""),
+                    enriched.get("rarity", ""),
+                    enriched.get("set_name", ""),
+                    enriched.get("edhrec_rank", 0),
+                    enriched.get("tcg_price", 0.0),
+                    enriched.get("salt_score", 0.0),
+                    enriched.get("is_game_changer", 0),
+                    enriched.get("category", "[]"),
+                    enriched.get("scryfall_id", ""),
+                    enriched.get("tcgplayer_id", ""),
+                    quantity,
+                    finish,
+                    condition,
+                    language,
+                    notes,
+                    tags,
+                    set_code,
+                    collector_number,
+                ),
+            )
+            conn.commit()
+            imported_count += 1
+
+    return {
+        "importedCount": imported_count,
+        "updatedCount": updated_count,
+        "failedCount": failed_count,
+        "errors": errors,
+    }
+
+
+# ══════════════════════════════════════════════════════════════
+# Scryfall Cache Management Endpoints
+# ══════════════════════════════════════════════════════════════
+
+
+@router.get("/api/cache/scryfall")
+async def scryfall_cache_stats():
+    """Return Scryfall response cache statistics."""
+    return _scryfall_cache.stats()
+
+
+@router.delete("/api/cache/scryfall")
+async def scryfall_cache_clear():
+    """Clear all cached Scryfall responses."""
+    deleted = _scryfall_cache.clear()
+    return {"cleared": deleted, "message": f"Deleted {deleted} cached entries"}
+
+
+@router.post("/api/cache/scryfall/evict")
+async def scryfall_cache_evict_expired():
+    """Remove only expired entries (older than TTL) from the cache."""
+    evicted = _scryfall_cache.evict_expired()
+    return {"evicted": evicted, "message": f"Evicted {evicted} expired entries"}
+
+

routes/deepseek.py

@@ -0,0 +1,800 @@
+"""
+routes/deepseek.py
+==================
+Python simulator & DeepSeek AI opponent endpoints:
+  POST /api/sim/run
+  GET  /api/sim/status
+  GET  /api/sim/result
+  POST /api/sim/run-from-deck
+  POST /api/sim/run-deepseek
+  POST /api/deepseek/connect
+  GET  /api/deepseek/status
+  POST /api/deepseek/configure
+  GET  /api/deepseek/logs
+"""
+from __future__ import annotations
+
+import json
+import logging
+import os
+import sys
+import time
+import threading as _threading
+import traceback
+import uuid as _uuid
+from pathlib import Path
+
+from fastapi import APIRouter, HTTPException
+from fastapi import Request as FastAPIRequest
+from fastapi.responses import JSONResponse
+
+from routes.shared import (
+    CFG, log_sim,
+    _get_db_conn,
+    _get_deepseek_brain,
+    _ml_logging_enabled,
+)
+from commander_ai_lab.sim.validator_brain import ValidatorBrain, ValidatorConfig
+logger = logging.getLogger(__name__)
+
+router = APIRouter(tags=["deepseek"])
+
+# In-memory store for simulation runs
+_sim_runs = {}  # sim_id -> { status, result, error }
+_sim_lock = _threading.Lock()
+
+# ── Validator singleton ──
+_validator_brain: ValidatorBrain | None = None
+_validator_lock = _threading.Lock()
+
+
+def _get_validator_brain() -> ValidatorBrain | None:
+    global _validator_brain
+    if os.environ.get("VALIDATOR_ENABLED", "false").lower() != "true":
+        return None
+    with _validator_lock:
+        if _validator_brain is None:
+            cfg = ValidatorConfig(
+                api_base=os.environ.get("VALIDATOR_API_BASE", "http://localhost:11434"),
+                model=os.environ.get("VALIDATOR_MODEL", "deepseek-r1:14b"),
+                max_tokens=int(os.environ.get("VALIDATOR_MAX_TOKENS", "2048")),
+                request_timeout=float(os.environ.get("VALIDATOR_TIMEOUT", "60.0")),
+            )
+            _validator_brain = ValidatorBrain(cfg)
+    return _validator_brain
+
+# ══════════════════════════════════════════════════════════════
+# Shared Helpers (extracted to deduplicate sim threads — #42)
+# ══════════════════════════════════════════════════════════════
+
+_SRC_PATH_ADDED = False
+
+
+def _ensure_src_path():
+    """Add project src/ to sys.path once."""
+    global _SRC_PATH_ADDED
+    if not _SRC_PATH_ADDED:
+        src_dir = str(Path(__file__).resolve().parent.parent / 'src')
+        if src_dir not in sys.path:
+            sys.path.insert(0, src_dir)
+        _SRC_PATH_ADDED = True
+
+
+def _build_deck_from_card_data(card_data: list[dict]) -> list:
+    """Build a list of Card objects from DB card dicts. Used by v2 + deepseek threads."""
+    _ensure_src_path()
+    from commander_ai_lab.sim.models import Card
+    from commander_ai_lab.sim.rules import enrich_card
+
+    deck = []
+    for cd in card_data:
+        c = Card(name=cd['name'])
+        if cd.get('type_line'):
+            c.type_line = cd['type_line']
+        if cd.get('cmc'):
+            c.cmc = float(cd['cmc'])
+        if cd.get('power') and cd.get('toughness'):
+            c.power = str(cd['power'])
+            c.toughness = str(cd['toughness'])
+            c.pt = c.power + '/' + c.toughness
+        if cd.get('oracle_text'):
+            c.oracle_text = cd['oracle_text']
+        if cd.get('mana_cost'):
+            c.mana_cost = cd['mana_cost']
+        if cd.get('keywords'):
+            kw = cd['keywords']
+            if isinstance(kw, str):
+                try:
+                    kw = json.loads(kw)
+                except Exception:
+                    kw = []
+            if isinstance(kw, list):
+                c.keywords = kw
+        enrich_card(c)
+        deck.append(c)
+    return deck
+
+
+def _run_game_loop(engine, deck_a, deck_b, deck_name, opponent_name,
+                   num_games, sim_id, **run_kwargs):
+    """Run *num_games* and accumulate stats. Returns (game_results, stats, elapsed)."""
+    start = time.time()
+    wins = losses = total_turns = 0
+    total_damage_dealt = total_damage_received = 0
+    total_spells_cast = total_creatures_played = 0
+    total_removal_used = total_ramp_played = 0
+    total_cards_drawn = total_max_board = 0
+    game_results = []
+
+    for i in range(num_games):
+        kw = dict(run_kwargs)
+        # DeepSeek engine needs per-game IDs
+        if 'game_id_prefix' in kw:
+            prefix = kw.pop('game_id_prefix')
+            kw['game_id'] = f'{prefix}-g{i+1}'
+        result = engine.run(deck_a, deck_b, name_a=deck_name,
+                            name_b=opponent_name, **kw)
+        game_data = result.to_dict()
+        game_data['gameNumber'] = i + 1
+              # Phase 8: Surface player_explanation from validation in turn log
+        for turn_entry in game_data.get('log', []):
+            v = turn_entry.get('validation')
+            if isinstance(v, dict) and v.get('player_explanation'):
+                turn_entry['player_explanation'] = v['player_explanation']
+        game_results.append(game_data)
+
+        if result.winner == 0:
+            wins += 1
+        else:
+            losses += 1
+        total_turns += result.turns
+
+        if result.player_a_stats:
+            s = result.player_a_stats
+            total_damage_dealt += s.damage_dealt
+            total_damage_received += s.damage_received
+            total_spells_cast += s.spells_cast
+            total_creatures_played += s.creatures_played
+            total_removal_used += s.removal_used
+            total_ramp_played += s.ramp_played
+            total_cards_drawn += s.cards_drawn
+            total_max_board += s.max_board_size
+
+        with _sim_lock:
+            _sim_runs[sim_id]['completed'] = i + 1
+
+    elapsed = time.time() - start
+    stats = dict(
+        wins=wins, losses=losses, total_turns=total_turns,
+        total_damage_dealt=total_damage_dealt,
+        total_damage_received=total_damage_received,
+        total_spells_cast=total_spells_cast,
+        total_creatures_played=total_creatures_played,
+        total_removal_used=total_removal_used,
+        total_ramp_played=total_ramp_played,
+        total_cards_drawn=total_cards_drawn,
+        total_max_board=total_max_board,
+    )
+    return game_results, stats, elapsed
+
+
+def _build_summary(stats: dict, deck_name: str, opponent_name: str,
+                   num_games: int, elapsed: float, **extra) -> dict:
+    """Build the JSON-serialisable summary dict from accumulated stats."""
+    n = num_games
+    summary = {
+        'deckName': deck_name,
+        'opponentName': opponent_name,
+        'totalGames': n,
+        'wins': stats['wins'],
+        'losses': stats['losses'],
+        'winRate': round(stats['wins'] / n * 100, 1) if n > 0 else 0.0,
+        'avgTurns': round(stats['total_turns'] / n, 1) if n > 0 else 0.0,
+        'avgDamageDealt': round(stats['total_damage_dealt'] / n, 1) if n > 0 else 0.0,
+        'avgDamageReceived': round(stats['total_damage_received'] / n, 1) if n > 0 else 0.0,
+        'avgSpellsCast': round(stats['total_spells_cast'] / n, 1) if n > 0 else 0.0,
+        'avgCreaturesPlayed': round(stats['total_creatures_played'] / n, 1) if n > 0 else 0.0,
+        'avgRemovalUsed': round(stats['total_removal_used'] / n, 1) if n > 0 else 0.0,
+        'avgRampPlayed': round(stats['total_ramp_played'] / n, 1) if n > 0 else 0.0,
+        'avgCardsDrawn': round(stats['total_cards_drawn'] / n, 1) if n > 0 else 0.0,
+        'avgMaxBoardSize': round(stats['total_max_board'] / n, 1) if n > 0 else 0.0,
+        'elapsedSeconds': round(elapsed, 3),
+    }
+    summary.update(extra)
+    return summary
+
+
+def _finish_sim(sim_id, summary, game_results):
+    """Mark a sim run as complete."""
+    with _sim_lock:
+        _sim_runs[sim_id]['status'] = 'complete'
+        _sim_runs[sim_id]['result'] = {'summary': summary, 'games': game_results}
+
+          # --- Persist batch-*.json for coach / training dashboard ---
+    try:
+        deck_name = summary.get('deckName', 'Unknown')
+        opponent_name = summary.get('opponentName', 'Training Deck')
+        batch_data = {
+            'format': 'python-sim',
+            'timestamp': time.strftime('%Y-%m-%dT%H:%M:%SZ', time.gmtime()),
+            'simId': sim_id,
+            'decks': [
+                {
+                    'deckName': deck_name,
+                    'seatIndex': 0,
+                    'commanderName': summary.get('commanderName', ''),
+                    'colorIdentity': summary.get('colorIdentity', []),
+                    'games': game_results,
+                },
+                {
+                    'deckName': opponent_name,
+                    'seatIndex': 1,
+                    'commanderName': '',
+                    'colorIdentity': [],
+                },
+            ],
+            'summary': summary,
+        }
+        results_dir = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'results')
+        os.makedirs(results_dir, exist_ok=True)
+        batch_path = os.path.join(results_dir, f'batch-sim-{sim_id}.json')
+        with open(batch_path, 'w') as f:
+            json.dump(batch_data, f, indent=2, default=str)
+        logger.info(f'Saved batch results to {batch_path}')
+
+        # Auto-generate deck report for training dashboard / coach
+        try:
+            from coach.report_generator import generate_deck_reports
+            reports_dir = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'deck-reports')
+            os.makedirs(reports_dir, exist_ok=True)
+            generate_deck_reports(str(results_dir), str(reports_dir))
+            logger.info(f'Generated deck reports after sim {sim_id}')
+        except ImportError:
+            logger.warning('reports module not available; skipping deck report generation')
+        except Exception as e:
+            logger.warning(f'Could not generate deck report: {e}')
+    except Exception as e:
+        logger.error(f'Failed to persist sim results: {e}')
+        
+def _fail_sim(sim_id, error):
+    """Mark a sim run as errored."""
+    with _sim_lock:
+        _sim_runs[sim_id]['status'] = 'error'
+        _sim_runs[sim_id]['error'] = str(error)
+
+# ══════════════════════════════════════════════════════════════
+# Background Sim Threads (now thin wrappers around shared helpers)
+# ══════════════════════════════════════════════════════════════
+
+def _run_sim_thread(sim_id: str, decklist: list, num_games: int,
+                    deck_name: str, record_logs: bool):
+    """Background thread: basic Monte Carlo sim from a raw decklist."""
+    try:
+        _ensure_src_path()
+        from commander_ai_lab.sim.engine import GameEngine
+        from commander_ai_lab.lab.experiments import build_deck, _generate_training_deck
+
+        with _sim_lock:
+            _sim_runs[sim_id]['status'] = 'running'
+
+        deck_a = build_deck(decklist)
+        deck_b = _generate_training_deck()
+        engine = GameEngine(max_turns=25, record_log=record_logs, ml_log=True)
+
+        game_results, stats, elapsed = _run_game_loop(
+            engine, deck_a, deck_b, deck_name, 'Training Deck',
+            num_games, sim_id)
+
+        summary = _build_summary(stats, deck_name, 'Training Deck',
+                                 num_games, elapsed)
+        _finish_sim(sim_id, summary, game_results)
+    except Exception as e:
+        _fail_sim(sim_id, e)
+        traceback.print_exc()
+
+
+def _run_sim_thread_v2(sim_id: str, card_data: list[dict], num_games: int,
+                       deck_name: str, record_logs: bool):
+    """Background thread: sim with full card data from the DB."""
+    try:
+        _ensure_src_path()
+        from commander_ai_lab.sim.engine import GameEngine
+        from commander_ai_lab.lab.experiments import _generate_training_deck
+
+        with _sim_lock:
+            _sim_runs[sim_id]['status'] = 'running'
+
+        deck_a = _build_deck_from_card_data(card_data)
+        deck_b = _generate_training_deck()
+        engine = GameEngine(max_turns=25, record_log=record_logs, ml_log=True)
+
+        game_results, stats, elapsed = _run_game_loop(
+            engine, deck_a, deck_b, deck_name, 'Training Deck',
+            num_games, sim_id)
+
+        summary = _build_summary(stats, deck_name, 'Training Deck',
+                                 num_games, elapsed)
+        _finish_sim(sim_id, summary, game_results)
+    except Exception as e:
+        _fail_sim(sim_id, e)
+        traceback.print_exc()
+
+
+def _run_sim_thread_deepseek(sim_id: str, card_data: list[dict],
+                             num_games: int, deck_name: str,
+                             record_logs: bool):
+    """Background thread: sim using DeepSeek AI opponent."""
+    try:
+        _ensure_src_path()
+        from commander_ai_lab.sim.deepseek_engine import DeepSeekGameEngine
+        from commander_ai_lab.lab.experiments import _generate_training_deck
+
+        with _sim_lock:
+            _sim_runs[sim_id]['status'] = 'running'
+
+        deck_a = _build_deck_from_card_data(card_data)
+        deck_b = _generate_training_deck()
+
+        brain = _get_deepseek_brain()
+        if brain and not brain._connected:
+            brain.check_connection()
+
+        validator = _get_validator_brain()  # None if disabled
+        engine = DeepSeekGameEngine(
+            brain=brain, ai_player_index=1,
+            max_turns=25, record_log=record_logs, ml_log=True,
+            validator=validator)
+
+        game_results, stats, elapsed = _run_game_loop(
+            engine, deck_a, deck_b, deck_name, 'DeepSeek AI',
+            num_games, sim_id,
+            game_id_prefix=f'ds-sim-{sim_id[:8]}',
+            archetype='midrange')
+
+        # Write ML decision JSONL for training pipeline
+        ml_decisions = engine.flush_ml_decisions()
+        if ml_decisions:
+            ml_path = os.path.join('results',
+                                   f'ml-decisions-sim-{sim_id[:8]}.jsonl')
+            os.makedirs('results', exist_ok=True)
+            with open(ml_path, 'w', encoding='utf-8') as mf:
+                for dec in ml_decisions:
+                    mf.write(json.dumps(dec) + '\n')
+
+        ds_stats = brain.get_stats() if brain else {}
+        summary = _build_summary(
+            stats, deck_name, 'DeepSeek AI', num_games, elapsed,
+            opponentType='deepseek', deepseekStats=ds_stats)
+        _finish_sim(sim_id, summary, game_results)
+    except Exception as e:
+        _fail_sim(sim_id, e)
+        traceback.print_exc()
+
+# ══════════════════════════════════════════════════════════════
+# N-Player Sim Thread
+# ══════════════════════════════════════════════════════════════
+
+
+def _run_sim_thread_n_player(sim_id: str, decks_data: list[list[dict]],
+                              deck_names: list[str], num_games: int,
+                              record_logs: bool, engine_type: str):
+    """Background thread: N-player (2-4) batch simulation."""
+    try:
+        _ensure_src_path()
+        from commander_ai_lab.sim.engine import GameEngine
+        from commander_ai_lab.sim.deepseek_engine import DeepSeekGameEngine
+
+        with _sim_lock:
+            _sim_runs[sim_id]['status'] = 'running'
+
+        n_players = len(decks_data)
+        built_decks = [_build_deck_from_card_data(cd) for cd in decks_data]
+
+        if engine_type == 'deepseek':
+            brain = _get_deepseek_brain()
+            if brain and not brain._connected:
+                brain.check_connection()
+            engine = DeepSeekGameEngine(
+                brain=brain, ai_player_index=1,
+                max_turns=25, record_log=record_logs, ml_log=True)
+        else:
+            engine = GameEngine(max_turns=25, record_log=record_logs, ml_log=True)
+
+        # Per-seat accumulators
+        seat_wins = [0] * n_players
+        seat_turns = [0] * n_players
+        seat_damage_dealt = [0] * n_players
+        seat_spells_cast = [0] * n_players
+        seat_creatures_played = [0] * n_players
+        seat_removal_used = [0] * n_players
+        seat_ramp_played = [0] * n_players
+        seat_cards_drawn = [0] * n_players
+        seat_max_board = [0] * n_players
+
+        game_results = []
+        start = time.time()
+
+        for i in range(num_games):
+            result = engine.run_n(decks=built_decks, names=deck_names)
+            game_data = result.to_dict()
+            game_data['gameNumber'] = i + 1
+            game_results.append(game_data)
+
+            if 0 <= result.winner_seat < n_players:
+                seat_wins[result.winner_seat] += 1
+
+            for pr in result.players:
+                si = pr.seat_index
+                seat_turns[si] += result.turns
+                if pr.stats:
+                    seat_damage_dealt[si] += pr.stats.damage_dealt
+                    seat_spells_cast[si] += pr.stats.spells_cast
+                    seat_creatures_played[si] += pr.stats.creatures_played
+                    seat_removal_used[si] += pr.stats.removal_used
+                    seat_ramp_played[si] += pr.stats.ramp_played
+                    seat_cards_drawn[si] += pr.stats.cards_drawn
+                    seat_max_board[si] += pr.stats.max_board_size
+
+            with _sim_lock:
+                _sim_runs[sim_id]['completed'] = i + 1
+
+        elapsed = time.time() - start
+        n = num_games if num_games > 0 else 1
+
+        players_summary = []
+        for si in range(n_players):
+            players_summary.append({
+                'seat': si,
+                'deckName': deck_names[si],
+                'wins': seat_wins[si],
+                'winRate': round(seat_wins[si] / n * 100, 1),
+                'avgTurns': round(seat_turns[si] / n, 1),
+                'avgDamageDealt': round(seat_damage_dealt[si] / n, 1),
+                'avgSpellsCast': round(seat_spells_cast[si] / n, 1),
+                'avgCreaturesPlayed': round(seat_creatures_played[si] / n, 1),
+                'avgRemovalUsed': round(seat_removal_used[si] / n, 1),
+                'avgRampPlayed': round(seat_ramp_played[si] / n, 1),
+                'avgCardsDrawn': round(seat_cards_drawn[si] / n, 1),
+                'avgMaxBoardSize': round(seat_max_board[si] / n, 1),
+            })
+
+        summary = {
+            'playerCount': n_players,
+            'totalGames': num_games,
+            'players': players_summary,
+            'elapsedSeconds': round(elapsed, 3),
+        }
+        _finish_sim(sim_id, summary, game_results)
+    except Exception as e:
+        _fail_sim(sim_id, e)
+        traceback.print_exc()
+
+
+# ══════════════════════════════════════════════════════════════
+# Simulation API Endpoints
+# ══════════════════════════════════════════════════════════════
+
+
+def _create_sim_run(num_games, deck_name):
+    """Create a queued sim run entry and return its ID."""
+    sim_id = str(_uuid.uuid4())[:8]
+    with _sim_lock:
+        _sim_runs[sim_id] = {
+            'status': 'queued', 'completed': 0,
+            'total': num_games, 'deckName': deck_name,
+            'result': None, 'error': None,
+        }
+    return sim_id
+
+
+def _fetch_deck_card_data(deck_id):
+    """Load card data from the DB for a given deck ID. Returns (deck_name, card_data) or raises."""
+    conn = _get_db_conn()
+    cur = conn.cursor()
+    cur.execute('SELECT name FROM decks WHERE id = ?', (deck_id,))
+    row = cur.fetchone()
+    if not row:
+        return None, None
+    deck_name = row[0]
+    cur.execute("""
+        SELECT dc.card_name, dc.quantity,
+               ce.type_line, ce.cmc, ce.power, ce.toughness,
+               ce.oracle_text, ce.keywords, ce.mana_cost
+        FROM deck_cards dc
+        LEFT JOIN (
+            SELECT scryfall_id, type_line, cmc, power, toughness,
+                   oracle_text, keywords, mana_cost
+            FROM collection_entries GROUP BY scryfall_id
+        ) ce ON ce.scryfall_id = dc.scryfall_id
+        WHERE dc.deck_id = ?
+    """, (deck_id,))
+    card_data = []
+    for r in cur.fetchall():
+        for _ in range(r[1] or 1):
+            card_data.append({
+                'name': r[0], 'type_line': r[2] or '',
+                'cmc': r[3] or 0, 'power': r[4] or '',
+                'toughness': r[5] or '', 'oracle_text': r[6] or '',
+                'keywords': r[7] or '', 'mana_cost': r[8] or '',
+            })
+    return deck_name, card_data
+
+
+@router.post('/api/sim/run')
+async def sim_run(request: FastAPIRequest):
+    """Start a Monte Carlo simulation with the Python engine."""
+    body = await request.json()
+    decklist = body.get('decklist', [])
+    num_games = body.get('numGames', 10)
+    deck_name = body.get('deckName', 'My Deck')
+    record_logs = body.get('recordLogs', True)
+
+    if not decklist:
+        return JSONResponse({'error': 'decklist is required'}, status_code=400)
+    if num_games < 1 or num_games > 1000:
+        return JSONResponse({'error': 'numGames must be 1-1000'}, status_code=400)
+
+    sim_id = _create_sim_run(num_games, deck_name)
+    _threading.Thread(target=_run_sim_thread,
+                      args=(sim_id, decklist, num_games, deck_name, record_logs),
+                      daemon=True).start()
+    return JSONResponse({'simId': sim_id, 'status': 'queued', 'total': num_games})
+
+
+@router.get('/api/sim/status')
+async def sim_status(simId: str):
+    """Poll simulation progress."""
+    with _sim_lock:
+        run = _sim_runs.get(simId)
+    if not run:
+        return JSONResponse({'error': 'sim not found'}, status_code=404)
+    return JSONResponse({
+        'simId': simId, 'status': run['status'],
+        'completed': run['completed'], 'total': run['total'],
+        'deckName': run.get('deckName', ''), 'error': run.get('error'),
+    })
+
+
+@router.get('/api/sim/result')
+async def sim_result(simId: str):
+    """Get completed simulation results."""
+    with _sim_lock:
+        run = _sim_runs.get(simId)
+    if not run:
+        return JSONResponse({'error': 'sim not found'}, status_code=404)
+    if run['status'] != 'complete':
+        return JSONResponse({'error': 'sim not complete', 'status': run['status']},
+                            status_code=400)
+    return JSONResponse(run['result'])
+
+
+@router.post('/api/sim/run-from-deck')
+async def sim_run_from_deck(request: FastAPIRequest):
+    """Start simulation using a deck from the Deck Builder (by deck ID)."""
+    body = await request.json()
+    deck_id = body.get('deckId')
+    num_games = body.get('numGames', 10)
+    record_logs = body.get('recordLogs', True)
+
+    if not deck_id:
+        return JSONResponse({'error': 'deckId required'}, status_code=400)
+
+    deck_name, card_data = _fetch_deck_card_data(deck_id)
+    if deck_name is None:
+        return JSONResponse({'error': 'deck not found'}, status_code=404)
+    if not card_data:
+        return JSONResponse({'error': 'deck has no cards'}, status_code=400)
+
+    sim_id = _create_sim_run(num_games, deck_name)
+    _threading.Thread(target=_run_sim_thread_v2,
+                      args=(sim_id, card_data, num_games, deck_name, record_logs),
+                      daemon=True).start()
+    return JSONResponse({'simId': sim_id, 'status': 'queued',
+                         'total': num_games, 'deckName': deck_name})
+
+
+@router.post('/api/sim/run-deepseek')
+async def sim_run_deepseek(request: FastAPIRequest):
+    """Start simulation using DeepSeek AI as the opponent brain."""
+    body = await request.json()
+    deck_id = body.get('deckId')
+    num_games = body.get('numGames', 5)
+    record_logs = body.get('recordLogs', True)
+
+    if not deck_id:
+        return JSONResponse({'error': 'deckId required'}, status_code=400)
+    if num_games < 1 or num_games > 50:
+        return JSONResponse({'error': 'numGames must be 1-50 for DeepSeek mode'},
+                            status_code=400)
+
+    deck_name, card_data = _fetch_deck_card_data(deck_id)
+    if deck_name is None:
+        return JSONResponse({'error': 'deck not found'}, status_code=404)
+    if not card_data:
+        return JSONResponse({'error': 'deck has no cards'}, status_code=400)
+
+    sim_id = _create_sim_run(num_games, deck_name)
+    _threading.Thread(target=_run_sim_thread_deepseek,
+                      args=(sim_id, card_data, num_games, deck_name, record_logs),
+                      daemon=True).start()
+    return JSONResponse({'simId': sim_id, 'status': 'queued',
+                         'total': num_games, 'deckName': deck_name,
+                         'opponentType': 'deepseek'})
+
+
+@router.post('/api/sim/run-n-player')
+async def sim_run_n_player(request: FastAPIRequest):
+    """Start an N-player (2-4) batch simulation."""
+    body = await request.json()
+    deck_ids = body.get('deckIds', [])
+    num_games = body.get('numGames', 10)
+    record_logs = body.get('recordLogs', True)
+    engine_type = body.get('engineType', 'heuristic')
+
+    if len(deck_ids) < 2 or len(deck_ids) > 4:
+        return JSONResponse({'error': '2-4 decks required'}, status_code=400)
+    if num_games < 1 or num_games > 1000:
+        return JSONResponse({'error': 'numGames must be 1-1000'}, status_code=400)
+
+    decks_data = []
+    deck_names = []
+    for did in deck_ids:
+        name, cards = _fetch_deck_card_data(did)
+        if name is None:
+            return JSONResponse({'error': f'Deck {did} not found'}, status_code=404)
+        if not cards:
+            return JSONResponse({'error': f'Deck {did} has no cards'}, status_code=400)
+        deck_names.append(name)
+        decks_data.append(cards)
+
+    sim_id = _create_sim_run(num_games, ' vs '.join(deck_names))
+    _threading.Thread(target=_run_sim_thread_n_player,
+                      args=(sim_id, decks_data, deck_names, num_games,
+                            record_logs, engine_type),
+                      daemon=True).start()
+    return JSONResponse({
+        'simId': sim_id, 'status': 'queued', 'total': num_games,
+        'deckNames': deck_names, 'playerCount': len(deck_ids),
+    })
+
+
+# ══════════════════════════════════════════════════════════════
+# DeepSeek AI Brain Endpoints
+# ══════════════════════════════════════════════════════════════
+
+
+@router.post('/api/deepseek/connect')
+async def deepseek_connect(request: FastAPIRequest):
+    """Test connection to the DeepSeek LLM endpoint and auto-detect model."""
+    body = await request.json() if await request.body() else {}
+    api_base = body.get('apiBase', None)
+    model = body.get('model', None)
+
+    brain = _get_deepseek_brain()
+    if brain is None:
+        return JSONResponse({'error': 'DeepSeek brain failed to initialize'},
+                            status_code=500)
+
+    if api_base:
+        brain.config.api_base = api_base
+    if model:
+        brain.config.model = model
+
+    connected = brain.check_connection()
+    return JSONResponse({
+        'connected': connected,
+        'apiBase': brain.config.api_base,
+        'model': brain.config.model,
+        'stats': brain.get_stats(),
+    })
+
+
+@router.get('/api/deepseek/status')
+async def deepseek_status():
+    """Get DeepSeek brain status and performance stats."""
+    brain = _get_deepseek_brain()
+    if brain is None:
+        return JSONResponse({'connected': False, 'error': 'Brain not initialized'})
+    return JSONResponse(brain.get_stats())
+
+
+@router.post('/api/deepseek/configure')
+async def deepseek_configure(request: FastAPIRequest):
+    """Update DeepSeek configuration at runtime."""
+    body = await request.json()
+    brain = _get_deepseek_brain()
+    if brain is None:
+        return JSONResponse({'error': 'Brain not initialized'}, status_code=500)
+
+    if 'apiBase' in body:
+        brain.config.api_base = body['apiBase']
+    if 'model' in body:
+        brain.config.model = body['model']
+    if 'temperature' in body:
+        brain.config.temperature = float(body['temperature'])
+    if 'maxTokens' in body:
+        brain.config.max_tokens = int(body['maxTokens'])
+    if 'timeout' in body:
+        brain.config.request_timeout = float(body['timeout'])
+    if 'cacheEnabled' in body:
+        brain.config.cache_enabled = bool(body['cacheEnabled'])
+    if 'logDecisions' in body:
+        brain.config.log_decisions = bool(body['logDecisions'])
+    if 'fallbackOnTimeout' in body:
+        brain.config.fallback_on_timeout = bool(body['fallbackOnTimeout'])
+
+    connected = brain.check_connection()
+    return JSONResponse({
+        'connected': connected,
+        'config': {
+            'apiBase': brain.config.api_base,
+            'model': brain.config.model,
+            'temperature': brain.config.temperature,
+            'maxTokens': brain.config.max_tokens,
+            'timeout': brain.config.request_timeout,
+            'cacheEnabled': brain.config.cache_enabled,
+            'logDecisions': brain.config.log_decisions,
+            'fallbackOnTimeout': brain.config.fallback_on_timeout,
+        },
+    })
+
+
+@router.get('/api/deepseek/logs')
+async def deepseek_logs():
+    """Get decision log stats and flush pending entries."""
+    brain = _get_deepseek_brain()
+    if brain is None:
+        return JSONResponse({'error': 'Brain not initialized'}, status_code=500)
+
+    pending = len(brain._decision_log)
+    flushed_path = None
+    if pending > 0:
+        try:
+            flushed_path = brain.flush_log()
+        except Exception as e:
+            return JSONResponse({'error': f'Failed to flush: {e}'},
+                                status_code=500)
+
+    import glob as _glob
+    log_dir = brain.config.log_dir
+    log_files = []
+    if log_dir and os.path.isdir(log_dir):
+        for f in sorted(_glob.glob(os.path.join(log_dir, 'decisions_*.jsonl'))):
+            stat = os.stat(f)
+            log_files.append({
+                'filename': os.path.basename(f),
+                'size_bytes': stat.st_size,
+                'modified': stat.st_mtime,
+            })
+
+    return JSONResponse({
+        'flushed': pending,
+        'flushedPath': flushed_path,
+        'logFiles': log_files[-20:],
+    })
+  
+
+@router.get("/api/validator/status")
+async def validator_status():
+    """Get DeepSeek-R1-14B validator brain status."""
+    v = _get_validator_brain()
+    if v is None:
+        return JSONResponse({"enabled": False,
+                             "reason": "VALIDATOR_ENABLED is not true"})
+    return JSONResponse({"enabled": True, **v.get_stats()})
+
+
+@router.post("/api/validator/configure")
+async def validator_configure(request: FastAPIRequest):
+    """Update validator configuration at runtime."""
+    body = await request.json()
+    v = _get_validator_brain()
+    if v is None:
+        return JSONResponse({"error": "Validator not enabled"}, status_code=400)
+    if "model" in body:
+        v.config.model = body["model"]
+    if "timeout" in body:
+        v.config.request_timeout = float(body["timeout"])
+    if "enabled" in body:
+        v.config.enabled = bool(body["enabled"])
+    return JSONResponse(v.get_stats())

routes/import_routes.py

@@ -0,0 +1,127 @@
+"""
+routes/import_routes.py
+=======================
+Deck import & commander meta endpoints:
+  POST /api/lab/import/url
+  POST /api/lab/import/text
+  GET  /api/lab/meta/commanders
+  GET  /api/lab/meta/search
+  POST /api/lab/meta/fetch
+"""
+from __future__ import annotations
+
+from typing import Optional
+
+from fastapi import APIRouter, HTTPException, BackgroundTasks
+from fastapi.responses import JSONResponse
+
+import routes.shared as _shared
+from routes.shared import (
+    ImportUrlRequest,
+    ImportTextRequest,
+    MetaFetchRequest,
+    _import_from_url,
+    _parse_text_decklist,
+    _save_profile_to_dck,
+    _fetch_edhrec_average,
+    log,
+)
+
+router = APIRouter(tags=["import"])
+
+
+@router.post("/api/lab/import/url")
+async def import_from_url(req: ImportUrlRequest):
+    """Import a deck from Archidekt or EDHREC URL. Returns parsed DeckProfile + saves .dck file."""
+    url = req.url.strip()
+    try:
+        profile = _import_from_url(url)
+        dck_path = _save_profile_to_dck(profile)
+        return {
+            "success": True,
+            "deckName": profile["name"],
+            "commander": profile["commander"],
+            "source": profile["source"],
+            "sourceUrl": profile["sourceUrl"],
+            "totalCards": profile["totalCards"],
+            "dckFile": dck_path.stem,
+            "colorIdentity": profile.get("colorIdentity", []),
+            "archetype": profile.get("archetype"),
+        }
+    except Exception as e:
+        raise HTTPException(400, str(e))
+
+
+@router.post("/api/lab/import/text")
+async def import_from_text(req: ImportTextRequest):
+    """Import a deck from plain text card list."""
+    try:
+        profile = _parse_text_decklist(req.text, req.commander)
+        dck_path = _save_profile_to_dck(profile)
+        return {
+            "success": True,
+            "deckName": profile["name"],
+            "commander": profile["commander"],
+            "source": "Text Import",
+            "totalCards": profile["totalCards"],
+            "dckFile": dck_path.stem,
+        }
+    except Exception as e:
+        raise HTTPException(400, str(e))
+
+
+@router.get("/api/lab/meta/commanders")
+async def list_meta_commanders():
+    """List available commanders in the meta mapping."""
+    commanders = []
+    for name, entries in _shared.COMMANDER_META.items():
+        entry = entries[0] if entries else {}
+        commanders.append({
+            "name": name,
+            "archetype": entry.get("archetype", ""),
+            "colorIdentity": entry.get("colorIdentity", []),
+            "source": entry.get("source", "edhrec"),
+        })
+    return {"commanders": commanders}
+
+
+@router.get("/api/lab/meta/search")
+async def search_meta_commanders(q: str = ""):
+    """Search commanders by partial name match."""
+    query = q.lower()
+    matches = []
+    for name, entries in _shared.COMMANDER_META.items():
+        if query in name.lower():
+            entry = entries[0] if entries else {}
+            matches.append({
+                "name": name,
+                "archetype": entry.get("archetype", ""),
+                "colorIdentity": entry.get("colorIdentity", []),
+            })
+    return {"results": matches}
+
+
+@router.post("/api/lab/meta/fetch")
+async def fetch_meta_deck(req: MetaFetchRequest):
+    """Fetch EDHREC average deck for a commander and save as .dck file."""
+    try:
+        profile = _fetch_edhrec_average(req.commander)
+        dck_path = _save_profile_to_dck(profile)
+        return {
+            "success": True,
+            "deckName": profile["name"],
+            "commander": profile["commander"],
+            "source": "EDHREC Average",
+            "sourceUrl": profile.get("sourceUrl", ""),
+            "totalCards": profile["totalCards"],
+            "dckFile": dck_path.stem,
+            "colorIdentity": profile.get("colorIdentity", []),
+            "sampleSize": profile.get("sampleSize"),
+        }
+    except Exception as e:
+        raise HTTPException(400, f"Failed to fetch meta deck for '{req.commander}': {str(e)}")
+
+
+# ══════════════════════════════════════════════════════════════
+# Existing Endpoints (v1/v2)
+# ══════════════════════════════════════════════════════════════

routes/lab.py

@@ -0,0 +1,461 @@
+"""
+routes/lab.py
+=============
+Batch simulation & lab management endpoints:
+  POST /api/lab/start
+  POST /api/lab/start-deepseek
+  GET  /api/lab/status
+  GET  /api/lab/result
+  GET  /api/lab/decks
+  GET  /api/lab/history
+  GET  /api/lab/profiles
+  GET  /api/lab/profiles/{name}
+  GET  /api/lab/analytics/{deck_name}
+  GET  /api/lab/trends/{deck_name}
+  GET  /api/lab/log
+  GET  /api/lab/debug-log
+"""
+from __future__ import annotations
+
+import glob
+import json
+import math
+import os
+import re
+import threading
+import uuid
+from datetime import datetime
+from pathlib import Path
+from typing import Optional
+
+from fastapi import APIRouter, HTTPException, BackgroundTasks, Request as FastAPIRequest
+from fastapi.responses import JSONResponse
+
+from routes.shared import (
+    CFG,
+    active_batches,
+    BatchState,
+    StartRequest,
+    StartResponse,
+    StatusResponse,
+    AI_PROFILES,
+    build_java_command,
+    run_batch_subprocess,
+    _run_deepseek_batch_thread,
+    parse_dck_file,
+    _get_db_conn,
+    _load_deck_cards_by_name,
+    _ml_logging_enabled,
+    log,
+    log_batch,
+)
+
+router = APIRouter(tags=["lab"])
+
+
+def _sanitize_floats(obj):
+    """Replace NaN/Infinity with None so JSONResponse (allow_nan=False) won't crash."""
+    if isinstance(obj, float):
+        if math.isnan(obj) or math.isinf(obj):
+            return None
+        return obj
+    if isinstance(obj, dict):
+        return {k: _sanitize_floats(v) for k, v in obj.items()}
+    if isinstance(obj, list):
+        return [_sanitize_floats(v) for v in obj]
+    return obj
+
+
+@router.post("/api/lab/start", response_model=StartResponse)
+async def start_batch(req: StartRequest, background_tasks: BackgroundTasks):
+    """Start a new batch simulation run."""
+    if len(req.decks) < 2 or len(req.decks) > 4:
+        raise HTTPException(400, "2-4 decks required")
+    if not all(req.decks):
+        raise HTTPException(400, f"All {len(req.decks)} deck slots must be filled")
+    if req.numGames < 1 or req.numGames > 10000:
+        raise HTTPException(400, "numGames must be 1-10000")
+    if req.threads < 1 or req.threads > 16:
+        raise HTTPException(400, "threads must be 1-16")
+
+    if not os.path.exists(CFG.forge_jar):
+        raise HTTPException(500, f"Forge JAR not found: {CFG.forge_jar}")
+    if not os.path.isdir(CFG.forge_dir):
+        raise HTTPException(500, f"Forge dir not found: {CFG.forge_dir}")
+
+    batch_id = str(uuid.uuid4())[:12]
+    state = BatchState(batch_id, req.numGames, req.threads)
+    active_batches[batch_id] = state
+
+    os.makedirs(CFG.results_dir, exist_ok=True)
+    output_path = os.path.join(CFG.results_dir, f"batch-{batch_id}.json")
+    state.result_path = output_path
+
+    background_tasks.add_task(
+        run_batch_subprocess,
+        state,
+        req.decks,
+        req.numGames,
+        req.threads,
+        req.seed,
+        req.clock,
+        output_path,
+        req.useLearnedPolicy,
+        req.policyStyle,
+        req.policyGreedy,
+        req.aiSimplified,
+        req.aiThinkTimeMs,
+        req.maxQueueDepth,
+    )
+
+    policy_msg = ""
+    if req.useLearnedPolicy:
+        policy_msg = f" [Learned Policy: {req.policyStyle}]"
+    return StartResponse(
+        batchId=batch_id,
+        status="started",
+        message=f"Running {req.numGames} games with {req.threads} threads{policy_msg}",
+    )
+
+
+@router.get("/api/lab/status", response_model=StatusResponse)
+async def get_status(batchId: Optional[str] = None):
+    # If no batchId given, find the most recent active (or last) batch
+    state = None
+    if batchId:
+        state = active_batches.get(batchId)
+    else:
+        # Return the most recently active batch, or any batch
+        for s in active_batches.values():
+            if s.running:
+                state = s
+                break
+        if state is None and active_batches:
+            state = list(active_batches.values())[-1]
+    if not state:
+        return StatusResponse(
+            batchId="", running=False, completed=0, total=0,
+            threads=0, elapsedMs=0, error=None, simsPerSec=0.0,
+            run_id="", games_completed=0, total_games=0, current_decks=[],
+        )
+    elapsed = int((datetime.now() - state.start_time).total_seconds() * 1000)
+    deck_names = getattr(state, 'deck_names', []) if hasattr(state, 'deck_names') else []
+    return StatusResponse(
+        batchId=state.batch_id,
+        running=state.running,
+        completed=state.completed_games,
+        total=state.total_games,
+        threads=state.threads,
+        elapsedMs=elapsed if state.running else state.elapsed_ms,
+        error=state.error,
+        simsPerSec=state.sims_per_sec,
+        # React SPA fields
+        run_id=state.batch_id,
+        games_completed=state.completed_games,
+        total_games=state.total_games,
+        current_decks=deck_names,
+    )
+
+
+@router.get("/api/lab/result")
+async def get_result(batchId: Optional[str] = None):
+    state = None
+    if batchId:
+        state = active_batches.get(batchId)
+    else:
+        # Find most recent completed batch
+        for s in reversed(list(active_batches.values())):
+            if not s.running and s.result_path:
+                state = s
+                break
+    if not state:
+        raise HTTPException(404, "No batch result available")
+    if state.running:
+        raise HTTPException(409, "Batch still running")
+    if state.error:
+        raise HTTPException(500, state.error)
+    if not state.result_path or not os.path.exists(state.result_path):
+        raise HTTPException(500, "Result file not found")
+    try:
+        with open(state.result_path, "r", encoding="utf-8") as f:
+            data = json.load(f)
+    except (json.JSONDecodeError, OSError) as e:
+        raise HTTPException(500, f"Failed to read result file: {e}")
+    data = _sanitize_floats(data)
+    return JSONResponse(content=data)
+
+
+@router.get("/api/lab/decks")
+async def list_decks():
+    decks = []
+    seen_names = set()
+
+    # 1. Forge .dck files from decks directory
+    decks_dir = CFG.forge_decks_dir
+    if decks_dir and os.path.isdir(decks_dir):
+        for f in sorted(Path(decks_dir).glob("*.dck")):
+            deck_name = f.stem
+            decks.append({"name": deck_name, "filename": f.name, "source": "forge"})
+            seen_names.add(deck_name.lower())
+
+    # 2. Decks from the Deck Builder database
+    try:
+        conn = _get_db_conn()
+        rows = conn.execute(
+            "SELECT id, name, commander_name FROM decks ORDER BY id DESC"
+        ).fetchall()
+        for row in rows:
+            db_name = row["name"]
+            if db_name.lower() not in seen_names:
+                decks.append({
+                    "name": db_name,
+                    "filename": "",
+                    "source": "deckbuilder",
+                    "deck_id": row["id"],
+                    "commander": row["commander_name"] or "",
+                })
+                seen_names.add(db_name.lower())
+    except Exception as e:
+        log.warning(f"  WARNING: Failed to load DB decks for /api/lab/decks: {e}")
+
+    return {"decks": decks}
+
+
+@router.get("/api/lab/history")
+async def list_history():
+    results_dir = Path(CFG.results_dir)
+    if not results_dir.exists():
+        return {"results": []}
+    results = []
+    for f in sorted(results_dir.glob("batch-*.json"), reverse=True):
+        try:
+            with open(f, "r", encoding="utf-8") as fh:
+                data = json.load(fh)
+            meta = data.get("metadata", {})
+            decks = data.get("decks", [])
+            results.append({
+                "filename": f.name,
+                "batchId": meta.get("batchId", ""),
+                "timestamp": meta.get("timestamp", ""),
+                "totalGames": meta.get("completedGames", 0),
+                "threads": meta.get("threads", 1),
+                "elapsedMs": meta.get("elapsedMs", 0),
+                "decks": [{"name": d.get("deckName", ""), "source": d.get("source", "")} for d in decks],
+            })
+        except Exception:
+            continue
+    return {"results": results[:50]}
+
+
+# ════════════════════════════════════════════════════════════
+# AI Profiles
+# ════════════════════════════════════════════════════════════
+
+AI_PROFILES = {
+    "default": {
+        "name": "default",
+        "description": "Balanced \u2014 Forge's default AI behavior",
+        "aggression": 0.5, "cardAdvantage": 0.5, "removalPriority": 0.5,
+        "boardPresence": 0.5, "comboPriority": 0.3, "patience": 0.5,
+    },
+    "aggro": {
+        "name": "aggro",
+        "description": "Aggressive \u2014 attacks early, prioritizes damage",
+        "aggression": 0.9, "cardAdvantage": 0.3, "removalPriority": 0.3,
+        "boardPresence": 0.8, "comboPriority": 0.1, "patience": 0.1,
+    },
+    "control": {
+        "name": "control",
+        "description": "Control \u2014 defensive, removal-heavy, card advantage",
+        "aggression": 0.2, "cardAdvantage": 0.9, "removalPriority": 0.9,
+        "boardPresence": 0.3, "comboPriority": 0.4, "patience": 0.9,
+    },
+    "combo": {
+        "name": "combo",
+        "description": "Combo \u2014 ramps, digs for pieces, assembles combos",
+        "aggression": 0.2, "cardAdvantage": 0.8, "removalPriority": 0.4,
+        "boardPresence": 0.3, "comboPriority": 0.95, "patience": 0.7,
+    },
+    "midrange": {
+        "name": "midrange",
+        "description": "Midrange \u2014 flexible, strong board presence, value-oriented",
+        "aggression": 0.5, "cardAdvantage": 0.6, "removalPriority": 0.6,
+        "boardPresence": 0.7, "comboPriority": 0.3, "patience": 0.5,
+    },
+}
+
+@router.get("/api/lab/profiles")
+async def list_profiles():
+    return {"profiles": list(AI_PROFILES.values())}
+
+@router.get("/api/lab/profiles/{name}")
+async def get_profile(name: str):
+    profile = AI_PROFILES.get(name.lower())
+    if not profile:
+        raise HTTPException(404, f"Profile '{name}' not found. Available: {list(AI_PROFILES.keys())}")
+    return profile
+
+
+@router.get("/api/lab/analytics/{deck_name}")
+async def analyze_deck(deck_name: str):
+    decks_dir = CFG.forge_decks_dir
+    if not decks_dir or not os.path.isdir(decks_dir):
+        raise HTTPException(500, f"Decks directory not found: {decks_dir}")
+    deck_path = os.path.join(decks_dir, deck_name + ".dck")
+    if not os.path.exists(deck_path):
+        for f in os.listdir(decks_dir):
+            if f.lower() == deck_name.lower() + ".dck":
+                deck_path = os.path.join(decks_dir, f)
+                break
+        else:
+            raise HTTPException(404, f"Deck not found: {deck_name}")
+    try:
+        analysis = parse_dck_file(deck_path)
+        return analysis
+    except Exception as e:
+        raise HTTPException(500, f"Analysis failed: {str(e)}")
+
+
+def parse_dck_file(deck_path: str) -> dict:
+    cards = []
+    commander = None
+    section = "Main"
+    deck_name = Path(deck_path).stem
+    with open(deck_path, "r", encoding="utf-8") as f:
+        for line in f:
+            line = line.strip()
+            if not line or line.startswith("//"):
+                continue
+            if line.startswith("[") and line.endswith("]"):
+                section = line[1:-1]
+                continue
+            if line.startswith("Name="):
+                deck_name = line[5:].strip()
+                continue
+            m = re.match(r"^(\d+)\s+(.+?)(?:\|(.+))?$", line)
+            if m:
+                qty = int(m.group(1))
+                name = m.group(2).strip()
+                set_code = m.group(3).strip() if m.group(3) else ""
+                cards.append({
+                    "quantity": qty, "name": name, "set": set_code,
+                    "section": section,
+                    "is_commander": 1 if section == "Commander" else 0,
+                })
+                if section == "Commander":
+                    commander = name
+    total = sum(c["quantity"] for c in cards)
+    return {
+        "deckName": deck_name,
+        "commanderName": commander,
+        "totalCards": total,
+        "cardCount": len(cards),
+        "cards": cards[:200],
+    }
+
+
+@router.get("/api/lab/trends/{deck_name}")
+async def get_deck_trends(deck_name: str):
+    results_dir = Path(CFG.results_dir)
+    if not results_dir.exists():
+        return {"deckName": deck_name, "history": []}
+    history = []
+    for f in sorted(results_dir.glob("batch-*.json")):
+        try:
+            with open(f, "r", encoding="utf-8") as fh:
+                data = json.load(fh)
+            summary = data.get("summary", {})
+            meta = data.get("metadata", {})
+            per_deck = summary.get("perDeck", [])
+            for ds in per_deck:
+                if ds.get("deckName", "").lower() == deck_name.lower():
+                    history.append({
+                        "batchId": meta.get("batchId"),
+                        "timestamp": meta.get("timestamp"),
+                        "winRate": ds.get("winRate", 0),
+                        "wins": ds.get("wins", 0),
+                        "losses": ds.get("losses", 0),
+                        "draws": ds.get("draws", 0),
+                        "totalGames": meta.get("completedGames", 0),
+                    })
+                    break
+        except Exception:
+            continue
+    return {"deckName": deck_name, "history": history}
+
+
+@router.get("/api/lab/log")
+async def get_log(batchId: str):
+    state = active_batches.get(batchId)
+    if not state:
+        raise HTTPException(404, f"Batch {batchId} not found")
+    return {"lines": state.log_lines[-200:]}
+
+
+@router.get("/api/lab/debug-log")
+async def get_debug_log():
+    """Return the raw Forge subprocess debug log for diagnosing simulation issues."""
+    # Check multiple possible locations for the debug log
+    candidates = []
+    if CFG.forge_dir:
+        candidates.append(Path(CFG.forge_dir).parent / "forge-sim-debug.log")
+        candidates.append(Path(CFG.forge_dir) / "forge-sim-debug.log")
+    candidates.append(Path("forge-sim-debug.log"))
+
+    for log_path in candidates:
+        if log_path.exists():
+            text = log_path.read_text(encoding="utf-8", errors="replace")
+            # Return last 50KB to avoid huge responses
+            if len(text) > 50000:
+                text = "... (truncated) ...\n" + text[-50000:]
+            return {"path": str(log_path), "content": text}
+
+    return {"path": None, "content": "No debug log found. Run a simulation first."}
+
+
+
+@router.post('/api/lab/start-deepseek')
+async def start_batch_deepseek(request: FastAPIRequest, background_tasks: BackgroundTasks):
+    """Start a batch simulation using the Python sim engine + DeepSeek LLM opponent."""
+    body = await request.json()
+    decks = body.get('decks', [])
+    num_games = body.get('numGames', 30)
+    # Note: threads not applicable for Python sim (single-threaded LLM calls)
+
+    if not decks or len(decks) < 1:
+        raise HTTPException(400, 'At least 1 deck required')
+    if num_games < 1 or num_games > 500:
+        raise HTTPException(400, 'numGames must be 1-500')
+
+    # Filter empty deck slots
+    decks = [d for d in decks if d]
+    if not decks:
+        raise HTTPException(400, 'No valid decks provided')
+
+    batch_id = str(uuid.uuid4())[:12]
+    state = BatchState(batch_id, num_games, 1)
+    active_batches[batch_id] = state
+
+    os.makedirs(CFG.results_dir, exist_ok=True)
+    output_path = os.path.join(CFG.results_dir, f'batch-{batch_id}.json')
+    state.result_path = output_path
+
+    # Use threading instead of BackgroundTasks to avoid async issues with DB
+    t = threading.Thread(
+        target=_run_deepseek_batch_thread,
+        args=(state, decks, num_games, output_path),
+        daemon=True,
+    )
+    t.start()
+
+    return JSONResponse({
+        'batchId': batch_id,
+        'status': 'started',
+        'message': f'Running {num_games} games across {len(decks)} decks with DeepSeek AI',
+        'engine': 'deepseek',
+    })
+
+
+# Global ML logging toggle (can be enabled via API)
+_ml_logging_enabled = True
+

routes/ml.py

@@ -0,0 +1,729 @@
+"""
+routes/ml.py
+============
+ML training data, policy inference, supervised training,
+PPO training, and tournament endpoints.
+
+  GET  /api/ml/status
+  GET  /api/ml/decisions/{filename}
+  GET  /api/ml/data/status
+  POST /api/ml/toggle
+  POST /api/ml/predict
+  POST /api/ml/predict/batch
+  GET  /api/ml/model
+  POST /api/ml/reload
+  POST /api/ml/train
+  GET  /api/ml/train/status
+  POST /api/ml/train/ppo
+  GET  /api/ml/train/ppo/status
+  POST /api/ml/tournament
+  GET  /api/ml/tournament/status
+  GET  /api/ml/tournament/results
+"""
+from __future__ import annotations
+
+import dataclasses
+import json
+import logging
+import os
+import threading
+from datetime import datetime
+from pathlib import Path
+
+from fastapi import APIRouter, HTTPException, Request as FastAPIRequest
+
+from routes.shared import CFG, _ml_logging_enabled
+
+log_ml = logging.getLogger("commander_ai_lab.ml")
+
+router = APIRouter(tags=["ml"])
+
+# ── module-level singletons ──────────────────────────────────────────────────
+_policy_service = None
+_policy_service_init_attempted = False
+
+# ── Thread-safe state objects ────────────────────────────────────────────────
+@dataclasses.dataclass
+class _TrainingState:
+    running: bool = False
+    progress: int = 0
+    total_epochs: int = 0
+    current_epoch: int = 0
+    phase: str = "idle"
+    message: str = ""
+    metrics: object = None
+    result: object = None
+    error: object = None
+    started_at: object = None
+
+    def snapshot(self) -> dict:
+        return dataclasses.asdict(self)
+
+@dataclasses.dataclass
+class _PPOState:
+    running: bool = False
+    iteration: int = 0
+    total_iterations: int = 0
+    phase: str = "idle"
+    message: str = ""
+    metrics: object = None
+    result: object = None
+    error: object = None
+
+    def snapshot(self) -> dict:
+        return dataclasses.asdict(self)
+
+@dataclasses.dataclass
+class _TournamentState:
+    running: bool = False
+    phase: str = "idle"
+    message: str = ""
+    result: object = None
+    error: object = None
+
+    def snapshot(self) -> dict:
+        return dataclasses.asdict(self)
+
+_training_state = _TrainingState()
+_training_lock  = threading.Lock()
+
+_ppo_state  = _PPOState()
+_ppo_lock   = threading.Lock()
+
+_tournament_state = _TournamentState()
+_tournament_lock  = threading.Lock()
+
+# module-level toggle (local copy; the shared one is for batch runs)
+_ml_logging_enabled_local = _ml_logging_enabled
+
+
+def _get_policy_service():
+    """Get or initialize the policy inference service (lazy loading)."""
+    global _policy_service, _policy_service_init_attempted
+    if _policy_service is None and not _policy_service_init_attempted:
+        _policy_service_init_attempted = True
+        try:
+            from ml.serving.policy_server import PolicyInferenceService
+            _policy_service = PolicyInferenceService()
+            if _policy_service.load():
+                log_ml.info(f"Policy model loaded on {_policy_service.device}")
+            else:
+                log_ml.error(f"Policy model not available: {_policy_service._load_error}")
+        except Exception as e:
+            log_ml.error(f"Policy service init failed: {e}")
+    return _policy_service
+
+
+# ==============================================================
+# ML Training Data Endpoints
+# ==============================================================
+
+@router.get("/api/ml/status")
+async def ml_status():
+    """Get ML decision logging status and available training data."""
+    import routes.shared as _shared
+    lab_root = Path(__file__).resolve().parent.parent
+    results_dir = lab_root / CFG.results_dir
+    ml_files = []
+    if results_dir.exists():
+        for f in sorted(results_dir.glob("ml-decisions-*.jsonl")):
+            lines = 0
+            try:
+                with open(f) as fh:
+                    lines = sum(1 for _ in fh)
+            except Exception:
+                pass
+            ml_files.append({
+                "file": f.name,
+                "decisions": lines,
+                "size_kb": round(f.stat().st_size / 1024, 1),
+            })
+    total_decisions = sum(f["decisions"] for f in ml_files)
+    return {
+        "ml_logging_enabled": _shared._ml_logging_enabled,
+        "training_files": ml_files,
+        "total_decisions": total_decisions,
+        "total_files": len(ml_files),
+    }
+
+
+@router.post("/api/ml/toggle")
+async def ml_toggle(enable: bool = True):
+    """Enable or disable ML decision logging for future batch runs."""
+    import routes.shared as _shared
+    _shared._ml_logging_enabled = enable
+    return {
+        "ml_logging_enabled": _shared._ml_logging_enabled,
+        "message": f"ML decision logging {'enabled' if enable else 'disabled'} for future batches",
+    }
+
+
+@router.get("/api/ml/decisions/{filename}")
+async def ml_get_decisions(filename: str, limit: int = 100, offset: int = 0):
+    """Read decision snapshots from a training data file."""
+    lab_root = Path(__file__).resolve().parent.parent
+    filepath = lab_root / CFG.results_dir / filename
+    if not filepath.exists() or not filename.startswith("ml-decisions-"):
+        raise HTTPException(404, f"ML decisions file not found: {filename}")
+    decisions = []
+    try:
+        with open(filepath) as f:
+            for i, line in enumerate(f):
+                if i < offset:
+                    continue
+                if len(decisions) >= limit:
+                    break
+                try:
+                    decisions.append(json.loads(line.strip()))
+                except json.JSONDecodeError:
+                    continue
+    except Exception as e:
+        raise HTTPException(500, f"Failed to read decisions: {e}")
+    return {
+        "file": filename,
+        "offset": offset,
+        "limit": limit,
+        "count": len(decisions),
+        "decisions": decisions,
+    }
+
+
+# ==============================================================
+# ML Policy Inference Endpoints
+# ==============================================================
+
+@router.post("/api/ml/predict")
+async def ml_predict(request: FastAPIRequest):
+    """Predict a macro-action from a game state snapshot."""
+    svc = _get_policy_service()
+    if svc is None or not svc._loaded:
+        detail = "Policy model not loaded. "
+        if svc:
+            detail += svc._load_error or "Train a model first."
+        else:
+            detail += "PyTorch may not be installed."
+        raise HTTPException(status_code=503, detail=detail)
+    body = await request.json()
+    playstyle = body.pop("archetype", "midrange")
+    temperature = body.pop("temperature", 1.0)
+    greedy = body.pop("greedy", False)
+    result = svc.predict(
+        decision_snapshot=body,
+        playstyle=playstyle,
+        temperature=temperature,
+        greedy=greedy,
+    )
+    if "error" in result:
+        raise HTTPException(status_code=500, detail=result["error"])
+    return result
+
+
+@router.post("/api/ml/predict/batch")
+async def ml_predict_batch(request: FastAPIRequest):
+    """Predict actions for multiple snapshots at once."""
+    svc = _get_policy_service()
+    if svc is None or not svc._loaded:
+        raise HTTPException(status_code=503, detail="Policy model not loaded")
+    body = await request.json()
+    snapshots = body.get("snapshots", [])
+    greedy = body.get("greedy", True)
+    if not snapshots:
+        return {"results": []}
+    results = svc.predict_batch(snapshots, greedy=greedy)
+    return {"results": results, "count": len(results)}
+
+
+@router.get("/api/ml/model")
+async def ml_model_info():
+    """Get information about the loaded policy model."""
+    svc = _get_policy_service()
+    if svc is None:
+        return {
+            "loaded": False,
+            "error": "Policy service not initialized",
+            "torch_available": False,
+        }
+    return svc.get_status()
+
+
+@router.post("/api/ml/reload")
+async def ml_reload_model(checkpoint: str = None):
+    """Hot-reload a policy model checkpoint."""
+    svc = _get_policy_service()
+    if svc is None:
+        raise HTTPException(status_code=503, detail="Policy service not available")
+    if not svc._loaded:
+        ok = svc.load()
+    else:
+        ok = svc.reload(checkpoint)
+    return {
+        "success": ok,
+        "status": svc.get_status(),
+    }
+
+
+# ==============================================================
+# ML Training Management Endpoints
+# ==============================================================
+
+def _run_training_pipeline(
+    results_dir: str, epochs: int, lr: float,
+    batch_size: int, patience: int, rebuild_dataset: bool,
+):
+    """Run the full ML training pipeline in a background thread."""
+    global _policy_service, _policy_service_init_attempted
+    try:
+        import sys
+        project_root = str(Path(__file__).resolve().parent.parent)
+        if project_root not in sys.path:
+            sys.path.insert(0, project_root)
+        with _training_lock:
+            _training_state.running = True
+            _training_state.started_at = datetime.now().isoformat()
+            _training_state.error = None
+            _training_state.result = None
+        data_dir = os.path.join(project_root, "ml", "models")
+        ckpt_dir = os.path.join(project_root, "ml", "models", "checkpoints")
+        os.makedirs(data_dir, exist_ok=True)
+        os.makedirs(ckpt_dir, exist_ok=True)
+        # Phase 1: Build Dataset
+        train_path = os.path.join(data_dir, "train.npz")
+        if rebuild_dataset or not os.path.exists(train_path):
+            with _training_lock:
+                _training_state.phase = "building"
+                _training_state.message = "Loading card embeddings & building dataset..."
+            log_ml.info("Building dataset (loading embeddings, may auto-download)...")
+            from ml.data.dataset_builder import build_dataset, split_dataset, save_dataset
+            dataset = build_dataset(results_dir=results_dir)
+            if not dataset:
+                raise RuntimeError("No training data produced. Check server log for details.")
+            train_ds, val_ds, test_ds = split_dataset(dataset)
+            save_dataset(train_ds, os.path.join(data_dir, "train.npz"))
+            save_dataset(val_ds, os.path.join(data_dir, "val.npz"))
+            save_dataset(test_ds, os.path.join(data_dir, "test.npz"))
+            total_samples = len(dataset["states"])
+            with _training_lock:
+                _training_state.message = f"Dataset built: {total_samples} samples"
+            log_ml.info(f"Dataset built: {total_samples} samples")
+        # Phase 2: Train
+        with _training_lock:
+            _training_state.phase = "training"
+            _training_state.total_epochs = epochs
+            _training_state.current_epoch = 0
+            _training_state.message = f"Training policy network ({epochs} epochs)..."
+        log_ml.info(f"Starting training: {epochs} epochs, lr={lr}, bs={batch_size}")
+        import numpy as np
+        import torch
+        from ml.training.policy_network import PolicyNetwork
+        from ml.training.trainer import SupervisedTrainer
+        def load_npz_split(path):
+            data = np.load(path)
+            return data["states"].astype(np.float32), data["labels"].astype(np.int64)
+        train_states, train_labels = load_npz_split(train_path)
+        val_path = os.path.join(data_dir, "val.npz")
+        val_states, val_labels = load_npz_split(val_path)
+        actual_dim = train_states.shape[1]
+        model = PolicyNetwork(input_dim=actual_dim)
+        device = "cpu"
+        if torch.cuda.is_available():
+            device = "cuda"
+        elif hasattr(torch.backends, "mps") and torch.backends.mps.is_available():
+            device = "mps"
+        trainer = SupervisedTrainer(
+            model=model, device=device, learning_rate=lr,
+            batch_size=batch_size, epochs=epochs, patience=patience,
+            checkpoint_dir=ckpt_dir,
+        )
+        summary = trainer.train(train_states, train_labels, val_states, val_labels)
+        with _training_lock:
+            _training_state.current_epoch = summary.get("epochs_trained", epochs)
+            _training_state.metrics = summary
+        # Phase 3: Evaluate
+        with _training_lock:
+            _training_state.phase = "evaluating"
+            _training_state.message = "Evaluating on test set..."
+        test_path = os.path.join(data_dir, "test.npz")
+        eval_results = None
+        if os.path.exists(test_path):
+            from ml.training.trainer import load_checkpoint, evaluate_model
+            test_states, test_labels = load_npz_split(test_path)
+            best_model, _ = load_checkpoint(summary["checkpoint_path"], device)
+            eval_results = evaluate_model(best_model, test_states, test_labels, device)
+            eval_path = os.path.join(ckpt_dir, "eval_results.json")
+            with open(eval_path, "w") as f:
+                json.dump(eval_results, f, indent=2)
+        # Done
+        with _training_lock:
+            _training_state.phase = "done"
+            _training_state.running = False
+            _training_state.result = {
+                "training": summary,
+                "evaluation": eval_results,
+                "checkpoint": summary.get("checkpoint_path", ""),
+                "device": device,
+            }
+            _training_state.message = f"Training complete! Best val acc: {summary.get('best_val_accuracy', 0):.1%}"
+        log_ml.info(f"Complete: {summary.get('best_val_accuracy', 0):.1%} val accuracy")
+        # Auto-reload policy server
+        if _policy_service and _policy_service._loaded:
+            _policy_service.reload(summary.get("checkpoint_path"))
+            log_ml.info("Policy server reloaded with new checkpoint")
+        elif not _policy_service_init_attempted:
+            _policy_service_init_attempted = False
+    except Exception as e:
+        import traceback
+        with _training_lock:
+            _training_state.phase = "error"
+            _training_state.running = False
+            _training_state.error = str(e)
+            _training_state.message = f"Training failed: {e}"
+        log_ml.error(f"ERROR: {e}")
+        traceback.print_exc()
+
+
+@router.post("/api/ml/train")
+async def ml_start_training(request: FastAPIRequest):
+    """Trigger ML training pipeline from the web UI."""
+    if _training_state.running:
+        raise HTTPException(409, "Training already in progress")
+    body = await request.json() if await request.body() else {}
+    epochs = body.get("epochs", 50)
+    lr = body.get("lr", 0.001)
+    batch_size = body.get("batchSize", 256)
+    patience = body.get("patience", 10)
+    rebuild = body.get("rebuildDataset", True)
+    results_dir = os.path.join(str(Path(__file__).resolve().parent.parent), "results")
+    with _training_lock:
+        _training_state.running = True
+        _training_state.progress = 0
+        _training_state.total_epochs = epochs
+        _training_state.current_epoch = 0
+        _training_state.phase = "starting"
+        _training_state.message = "Initializing training pipeline..."
+        _training_state.metrics = None
+        _training_state.result = None
+        _training_state.error = None
+    thread = threading.Thread(
+        target=_run_training_pipeline,
+        args=(results_dir, epochs, lr, batch_size, patience, rebuild),
+        daemon=True,
+    )
+    thread.start()
+    return {
+        "status": "started",
+        "config": {
+            "epochs": epochs, "lr": lr, "batchSize": batch_size,
+            "patience": patience, "rebuildDataset": rebuild,
+        },
+    }
+
+
+@router.get("/api/ml/train")
+async def ml_training_status_poll():
+    """Return current training status (polled by Unity client)."""
+    with _training_lock:
+        return _training_state.snapshot()
+
+
+@router.get("/api/ml/train/status")
+async def ml_training_status():
+    """Get current training pipeline status."""
+    with _training_lock:
+        return _training_state.snapshot()
+
+
+@router.get("/api/ml/data/status")
+async def ml_data_status():
+    """Get status of available training data and model checkpoints."""
+    project_root = Path(__file__).resolve().parent.parent
+    results_dir = project_root / "results"
+    data_dir = project_root / "ml" / "models"
+    ckpt_dir = data_dir / "checkpoints"
+    decision_files = []
+    if results_dir.exists():
+        for f in sorted(results_dir.glob("ml-decisions-*.jsonl")):
+            line_count = 0
+            try:
+                with open(f, "r") as fh:
+                    line_count = sum(1 for _ in fh)
+            except Exception:
+                pass
+            decision_files.append({"name": f.name, "size": f.stat().st_size, "decisions": line_count})
+    datasets = {}
+    for split in ["train", "val", "test"]:
+        path = data_dir / f"{split}.npz"
+        if path.exists():
+            try:
+                import numpy as np
+                data = np.load(str(path))
+                datasets[split] = {
+                    "samples": int(data["states"].shape[0]),
+                    "features": int(data["states"].shape[1]),
+                    "size": path.stat().st_size,
+                }
+            except Exception:
+                datasets[split] = {"size": path.stat().st_size}
+    checkpoints = []
+    if ckpt_dir.exists():
+        for f in sorted(ckpt_dir.glob("*.pt")):
+            checkpoints.append({
+                "name": f.name,
+                "size": f.stat().st_size,
+                "modified": datetime.fromtimestamp(f.stat().st_mtime).isoformat(),
+            })
+    eval_path = ckpt_dir / "eval_results.json"
+    eval_results = None
+    if eval_path.exists():
+        try:
+            with open(eval_path, "r") as f:
+                eval_results = json.load(f)
+        except Exception:
+            pass
+    return {
+        "decisionFiles": decision_files,
+        "totalDecisions": sum(d["decisions"] for d in decision_files),
+        "datasets": datasets,
+        "checkpoints": checkpoints,
+        "evalResults": eval_results,
+        "policyLoaded": _policy_service is not None and _policy_service._loaded if _policy_service else False,
+    }
+
+
+# ==============================================================
+# PPO Training + Tournament Endpoints
+# ==============================================================
+
+def _run_ppo_pipeline(
+    iterations: int, episodes_per_iter: int, ppo_epochs: int,
+    batch_size: int, lr: float, clip_epsilon: float,
+    entropy_coeff: float, opponent: str, playstyle: str,
+    load_supervised: str,
+):
+    """Run PPO training in a background thread."""
+    try:
+        import sys
+        project_root = str(Path(__file__).resolve().parent.parent)
+        if project_root not in sys.path:
+            sys.path.insert(0, project_root)
+        with _ppo_lock:
+            _ppo_state.running = True
+            _ppo_state.error = None
+            _ppo_state.result = None
+        from ml.training.ppo_trainer import PPOTrainer, PPOConfig
+        ckpt_dir = os.path.join(project_root, "ml", "models", "checkpoints")
+        config = PPOConfig(
+            iterations=iterations, episodes_per_iter=episodes_per_iter,
+            ppo_epochs=ppo_epochs, batch_size=batch_size,
+            learning_rate=lr, clip_epsilon=clip_epsilon,
+            entropy_coeff=entropy_coeff, opponent=opponent,
+            playstyle=playstyle, checkpoint_dir=ckpt_dir,
+            load_supervised=load_supervised if load_supervised else None,
+        )
+        trainer = PPOTrainer(config)
+        def progress_cb(iteration, metrics):
+            with _ppo_lock:
+                _ppo_state.iteration = iteration
+                _ppo_state.phase = "training"
+                _ppo_state.message = f"Iteration {iteration}/{iterations} | WR: {metrics.get('win_rate', 0):.0%}"
+                _ppo_state.metrics = metrics
+        summary = trainer.train(progress_callback=progress_cb)
+        with _ppo_lock:
+            _ppo_state.phase = "done"
+            _ppo_state.running = False
+            _ppo_state.result = summary
+            _ppo_state.message = f"PPO complete! Best win rate: {summary.get('best_win_rate', 0):.0%}"
+    except Exception as e:
+        import traceback
+        with _ppo_lock:
+            _ppo_state.phase = "error"
+            _ppo_state.running = False
+            _ppo_state.error = str(e)
+            _ppo_state.message = f"PPO failed: {e}"
+        traceback.print_exc()
+
+
+@router.post("/api/ml/train/ppo")
+async def ml_start_ppo(request: FastAPIRequest):
+    """Start PPO training pipeline."""
+    if _ppo_state.running:
+        raise HTTPException(409, "PPO training already in progress")
+    if _training_state.running:
+        raise HTTPException(409, "Supervised training in progress -- wait for it to finish")
+    body = await request.json() if await request.body() else {}
+    iterations = body.get("iterations", 100)
+    episodes = body.get("episodesPerIter", 64)
+    ppo_epochs = body.get("ppoEpochs", 4)
+    batch_size = body.get("batchSize", 256)
+    lr = body.get("lr", 3e-4)
+    clip_eps = body.get("clipEpsilon", 0.2)
+    entropy = body.get("entropyCoeff", 0.01)
+    opponent = body.get("opponent", "heuristic")
+    playstyle = body.get("playstyle", "midrange")
+    load_sup = body.get("loadSupervised", "")
+    with _ppo_lock:
+        _ppo_state.running = True
+        _ppo_state.iteration = 0
+        _ppo_state.total_iterations = iterations
+        _ppo_state.phase = "starting"
+        _ppo_state.message = "Initializing PPO..."
+        _ppo_state.metrics = None
+        _ppo_state.result = None
+        _ppo_state.error = None
+    thread = threading.Thread(
+        target=_run_ppo_pipeline,
+        args=(iterations, episodes, ppo_epochs, batch_size, lr,
+              clip_eps, entropy, opponent, playstyle, load_sup),
+        daemon=True,
+    )
+    thread.start()
+    return {"status": "started", "iterations": iterations}
+
+
+@router.get("/api/ml/train/ppo/status")
+async def ml_ppo_status():
+    """Get PPO training status."""
+    with _ppo_lock:
+        return _ppo_state.snapshot()
+
+
+def _run_tournament_pipeline(episodes: int, playstyle: str):
+    """Run tournament in a background thread."""
+    try:
+        import sys
+        project_root = str(Path(__file__).resolve().parent.parent)
+        if project_root not in sys.path:
+            sys.path.insert(0, project_root)
+        with _tournament_lock:
+            _tournament_state.running = True
+            _tournament_state.error = None
+            _tournament_state.result = None
+            _tournament_state.phase = "running"
+            _tournament_state.message = "Running tournament..."
+        from ml.eval.tournament import (
+            run_tournament, HeuristicPolicy, RandomPolicy, LearnedPolicy,
+        )
+        ckpt_dir = os.path.join(project_root, "ml", "models", "checkpoints")
+        policies = {
+            "heuristic": HeuristicPolicy(),
+            "random": RandomPolicy(),
+        }
+        sup_path = os.path.join(ckpt_dir, "best_policy.pt")
+        if os.path.exists(sup_path):
+            policies["supervised"] = LearnedPolicy(sup_path)
+        ppo_path = os.path.join(ckpt_dir, "best_ppo.pt")
+        if os.path.exists(ppo_path):
+            policies["ppo"] = LearnedPolicy(ppo_path)
+        result = run_tournament(
+            policies=policies,
+            episodes_per_matchup=episodes,
+            playstyle=playstyle,
+        )
+        output_path = os.path.join(ckpt_dir, "tournament_results.json")
+        with open(output_path, "w") as f:
+            json.dump(result.to_dict(), f, indent=2)
+        with _tournament_lock:
+            _tournament_state.phase = "done"
+            _tournament_state.running = False
+            _tournament_state.result = result.to_dict()
+            _tournament_state.message = f"Tournament complete! {result.total_matches} matches"
+    except Exception as e:
+        import traceback
+        with _tournament_lock:
+            _tournament_state.phase = "error"
+            _tournament_state.running = False
+            _tournament_state.error = str(e)
+            _tournament_state.message = f"Tournament failed: {e}"
+        traceback.print_exc()
+
+
+@router.post("/api/ml/tournament")
+async def ml_start_tournament(request: FastAPIRequest):
+    """Start a tournament evaluation."""
+    if _tournament_state.running:
+        raise HTTPException(409, "Tournament already in progress")
+    body = await request.json() if await request.body() else {}
+    episodes = body.get("episodes", 50)
+    playstyle = body.get("playstyle", "midrange")
+    with _tournament_lock:
+        _tournament_state.running = True
+        _tournament_state.phase = "starting"
+        _tournament_state.message = "Initializing tournament..."
+        _tournament_state.result = None
+        _tournament_state.error = None
+    thread = threading.Thread(
+        target=_run_tournament_pipeline,
+        args=(episodes, playstyle),
+        daemon=True,
+    )
+    thread.start()
+    return {"status": "started", "episodes": episodes}
+
+
+@router.get("/api/ml/tournament/status")
+async def ml_tournament_status():
+    """Get tournament status."""
+    with _tournament_lock:
+        return _tournament_state.snapshot()
+
+
+@router.get("/api/ml/tournament/results")
+async def ml_tournament_results():
+    """Get latest tournament results."""
+    project_root = Path(__file__).resolve().parent.parent
+    results_path = project_root / "ml" / "models" / "checkpoints" / "tournament_results.json"
+    if results_path.exists():
+        with open(results_path, "r") as f:
+            return json.load(f)
+    return {"error": "No tournament results found. Run a tournament first."}
+
+
+# ==============================================================
+# Model listing and stats endpoints
+# ==============================================================
+
+@router.get("/api/ml/models")
+async def ml_list_models():
+    """List available policy model checkpoints."""
+    project_root = Path(__file__).resolve().parent.parent
+    ckpt_dir = project_root / "ml" / "models" / "checkpoints"
+    checkpoints = []
+    if ckpt_dir.exists():
+        for f in sorted(ckpt_dir.glob("*.pt")):
+            checkpoints.append({
+                "name": f.name,
+                "path": str(f),
+                "size_kb": round(f.stat().st_size / 1024, 1),
+                "modified": datetime.fromtimestamp(f.stat().st_mtime).isoformat(),
+            })
+    svc = _get_policy_service()
+    return {
+        "models": checkpoints,
+        "count": len(checkpoints),
+        "active_model": svc.get_status() if svc else None,
+    }
+
+
+@router.get("/api/ml/stats")
+async def ml_stats():
+    """Return aggregate ML training statistics and model performance metrics."""
+    project_root = Path(__file__).resolve().parent.parent
+    ckpt_dir = project_root / "ml" / "models" / "checkpoints"
+    eval_results = None
+    eval_path = ckpt_dir / "eval_results.json"
+    if eval_path.exists():
+        try:
+            with open(eval_path, "r") as f:
+                eval_results = json.load(f)
+        except Exception:
+            pass
+    with _training_lock:
+        training_snapshot = _training_state.snapshot()
+    with _ppo_lock:
+        ppo_snapshot = _ppo_state.snapshot()
+    svc = _get_policy_service()
+    return {
+        "training": training_snapshot,
+        "ppo": ppo_snapshot,
+        "eval_results": eval_results,
+        "policy_loaded": svc._loaded if svc else False,
+    }

routes/precon.py

@@ -0,0 +1,96 @@
+"""
+routes/precon.py
+================
+Precon deck management endpoints:
+  GET  /api/lab/precons
+  POST /api/lab/precons/install
+  POST /api/lab/precons/install-batch
+  POST /api/lab/precons/refresh
+"""
+from __future__ import annotations
+
+import os
+import shutil
+from pathlib import Path
+
+from fastapi import APIRouter, HTTPException
+
+import routes.shared as _shared                    # module reference, not value import
+from routes.shared import CFG, download_precon_database, log
+from services.precon_service import _get_precon_dir
+
+router = APIRouter(tags=["precon"])
+
+
+@router.get("/api/lab/precons")
+async def list_precons():
+    """List all available precon decks."""
+    return {"precons": _shared.PRECON_INDEX}        # reads current value every call
+
+
+@router.post("/api/lab/precons/install")
+async def install_precon(req: dict):
+    """Install a precon deck to the Forge decks directory.
+    Body: {"fileName": "Elven_Empire.dck"} or {"name": "Elven_Empire.dck"}
+    """
+    file_name = req.get("fileName", "") or req.get("name", "")
+    if not file_name:
+        raise HTTPException(400, "fileName is required")
+
+    src = _get_precon_dir() / file_name
+    if not src.exists():
+        raise HTTPException(404, f"Precon not found: {file_name}")
+
+    decks_dir = CFG.forge_decks_dir
+    if not decks_dir or not os.path.isdir(decks_dir):
+        raise HTTPException(500, f"Forge decks directory not found: {decks_dir}")
+
+    dst = Path(decks_dir) / file_name
+    shutil.copy2(str(src), str(dst))
+
+    deck_name = file_name.replace(".dck", "")
+    return {
+        "installed": True,
+        "deckName": deck_name,
+        "destination": str(dst),
+        "message": f"Installed {deck_name} to Forge decks",
+    }
+
+
+@router.post("/api/lab/precons/install-batch")
+async def install_precons_batch(req: dict):
+    """Install multiple precon decks at once.
+    Body: {"fileNames": ["Elven_Empire.dck", ...]} or {"names": ["Elven_Empire.dck", ...]}
+    """
+    file_names = req.get("fileNames", []) or req.get("names", [])
+    if not file_names:
+        raise HTTPException(400, "fileNames list is required")
+
+    decks_dir = CFG.forge_decks_dir
+    if not decks_dir or not os.path.isdir(decks_dir):
+        raise HTTPException(500, f"Forge decks directory not found: {decks_dir}")
+
+    results = []
+    for file_name in file_names:
+        src = _get_precon_dir() / file_name
+        if not src.exists():
+            results.append({"fileName": file_name, "installed": False, "error": "not found"})
+            continue
+        dst = Path(decks_dir) / file_name
+        shutil.copy2(str(src), str(dst))
+        deck_name = file_name.replace(".dck", "")
+        results.append({"fileName": file_name, "installed": True, "deckName": deck_name})
+
+    return {"results": results}
+
+
+@router.post("/api/lab/precons/refresh")
+async def refresh_precons():
+    """Force re-download of all Commander precon decks from GitHub."""
+    result = download_precon_database(force=True)
+    if result.get("error"):
+        raise HTTPException(502, result["error"])
+    return {
+        "message": f"Downloaded {result['downloaded']} Commander precon decks",
+        "total": result.get("total", 0),
+    }

routes/scanner.py

@@ -0,0 +1,481 @@
+"""
+routes/scanner.py
+=================
+Card scanner & collection enrichment endpoints:
+  POST /api/collection/scan
+  POST /api/collection/scan/add
+  POST /api/collection/re-enrich
+  POST /api/collection/auto-classify
+  POST /api/collection/auto-classify-all
+  GET  /api/collection/{cardId}/edhrec
+"""
+from __future__ import annotations
+
+import json
+import os
+import re
+import time
+from pathlib import Path
+from urllib.request import urlopen, Request
+from urllib.error import URLError
+
+from fastapi import APIRouter, HTTPException
+from fastapi import Request as FastAPIRequest
+
+from routes.shared import (
+    CFG, log_scan, log_deckgen,
+    _get_db_conn, _row_to_dict, _enrich_from_scryfall,
+    _scryfall_fuzzy_lookup, _detect_card_roles,
+    _to_edhrec_slug, _API_HEADERS,
+    _edhrec_cache_get, _edhrec_cache_set,
+)
+
+router = APIRouter(tags=["scanner"])
+
+
+@router.post("/api/collection/scan")
+async def scan_card_image(request: FastAPIRequest):
+    """
+    Scan a card image and return recognized card name(s).
+
+    Uses Ximilar Visual AI for card recognition.
+
+    Accepts multipart/form-data with:
+      - file: image file (JPEG/PNG)
+      - mode: "single" (default) or "multi"
+
+    Returns:
+      { "results": [ { raw_ocr, matched_name, set_code, scryfall_id, confidence, image_uri, error, collector_number, rarity, tcgplayer_url }, ... ] }
+    """
+    try:
+        from scanner.pipeline import scan_single, scan_multi
+    except ImportError as e:
+        raise HTTPException(500, f"Scanner module not available: {e}")
+
+    # Try CFG first, then fall back to env var
+    api_key = CFG.ximilar_api_key or os.environ.get("XIMILAR_API_KEY", "")
+    if not api_key:
+        raise HTTPException(500, "Card scanner not configured. Start the server with --ximilar-key YOUR_KEY or set XIMILAR_API_KEY env var.")
+
+    # Parse multipart form
+    form = await request.form()
+    upload = form.get("file")
+    if upload is None:
+        raise HTTPException(400, "No file uploaded. Send an image as 'file' in multipart/form-data.")
+
+    mode = str(form.get("mode", "single")).lower()
+    image_bytes = await upload.read()
+
+    if len(image_bytes) == 0:
+        raise HTTPException(400, "Uploaded file is empty")
+
+    # Limit file size (20 MB)
+    if len(image_bytes) > 20 * 1024 * 1024:
+        raise HTTPException(400, "Image file too large (max 20 MB)")
+
+    log_scan.info(f"  Processing {len(image_bytes)} bytes, mode={mode} (Ximilar AI)")
+
+    if mode == "multi":
+        results = scan_multi(image_bytes, _scryfall_fuzzy_lookup, ximilar_api_key=api_key)
+    else:
+        result = scan_single(image_bytes, _scryfall_fuzzy_lookup, ximilar_api_key=api_key)
+        results = [result]
+
+    return {
+        "results": [r.to_dict() for r in results]
+    }
+
+
+@router.post("/api/collection/scan/add")
+async def scan_add_cards(body: dict):
+    """
+    Add scanned cards to the collection.
+
+    Body:
+      {
+        "cards": [
+          { "name": "Lightning Bolt", "set_code": "m11", "quantity": 1 },
+          ...
+        ]
+      }
+    """
+    cards = body.get("cards", [])
+    if not cards:
+        raise HTTPException(400, "No cards provided")
+
+    conn = _get_db_conn()
+    imported = 0
+    updated = 0
+    errors = []
+
+    for card_req in cards:
+        name = str(card_req.get("name", "")).strip()
+        if not name:
+            continue
+
+        quantity = int(card_req.get("quantity", 1))
+        set_code = str(card_req.get("set_code", "")).lower()
+        collector_number = str(card_req.get("collector_number", ""))
+        finish = "NORMAL"
+
+        try:
+            enriched = _enrich_from_scryfall(name, set_code, collector_number)
+        except Exception as e:
+            errors.append(f"Scryfall error for '{name}': {e}")
+            continue
+
+        if not enriched or "_error" in enriched:
+            err_msg = enriched.get("_error", f"Not found: '{name}'") if enriched else f"Not found: '{name}'"
+            errors.append(err_msg)
+            continue
+
+        resolved_name = enriched.get("name", name)
+
+        # Check for existing entry
+        existing = conn.execute(
+            """SELECT id, quantity FROM collection_entries
+               WHERE name = ? AND set_code = ? AND collector_number = ? AND finish = ?""",
+            (resolved_name, set_code, collector_number, finish),
+        ).fetchone()
+
+        if existing:
+            new_qty = existing["quantity"] + quantity
+            conn.execute(
+                "UPDATE collection_entries SET quantity = ?, updated_at = datetime('now') WHERE id = ?",
+                (new_qty, existing["id"]),
+            )
+            conn.commit()
+            updated += 1
+        else:
+            conn.execute(
+                """INSERT INTO collection_entries
+                   (name, type_line, subtypes, is_legendary, is_basic, color_identity,
+                    cmc, mana_cost, oracle_text, keywords,
+                    power, toughness, rarity, set_name, edhrec_rank,
+                    tcg_price, salt_score, is_game_changer,
+                    category, scryfall_id, tcgplayer_id,
+                    quantity, finish, condition, language, notes, tags,
+                    set_code, collector_number)
+                   VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)""",
+                (
+                    resolved_name,
+                    enriched.get("type_line", ""),
+                    enriched.get("subtypes", "[]"),
+                    enriched.get("is_legendary", 0),
+                    enriched.get("is_basic", 0),
+                    enriched.get("color_identity", "[]"),
+                    enriched.get("cmc", 0.0),
+                    enriched.get("mana_cost", ""),
+                    enriched.get("oracle_text", ""),
+                    enriched.get("keywords", "[]"),
+                    enriched.get("power", ""),
+                    enriched.get("toughness", ""),
+                    enriched.get("rarity", ""),
+                    enriched.get("set_name", ""),
+                    enriched.get("edhrec_rank", 0),
+                    enriched.get("tcg_price", 0.0),
+                    enriched.get("salt_score", 0.0),
+                    enriched.get("is_game_changer", 0),
+                    enriched.get("category", "[]"),
+                    enriched.get("scryfall_id", ""),
+                    enriched.get("tcgplayer_id", ""),
+                    quantity,
+                    finish,
+                    "",  # condition
+                    "",  # language
+                    "",  # notes
+                    "",  # tags
+                    set_code,
+                    collector_number,
+                ),
+            )
+            conn.commit()
+            imported += 1
+
+    return {
+        "importedCount": imported,
+        "updatedCount": updated,
+        "failedCount": len(errors),
+        "errors": errors,
+    }
+
+
+# ══════════════════════════════════════════════════════════════
+# Re-enrich Collection from Scryfall
+# ══════════════════════════════════════════════════════════════
+
+@router.post("/api/collection/re-enrich")
+async def re_enrich_collection():
+    """
+    Re-fetch Scryfall data for all collection entries to backfill
+    missing fields (mana_cost, power, toughness, rarity, set_name, edhrec_rank).
+    Also refreshes: type_line, oracle_text, keywords, cmc, color_identity, prices.
+    Runs synchronously — may take a while for large collections.
+    """
+    conn = _get_db_conn()
+    rows = conn.execute(
+        "SELECT id, name, set_code, collector_number FROM collection_entries ORDER BY id"
+    ).fetchall()
+
+    total = len(rows)
+    enriched_count = 0
+    skipped_count = 0
+    errors = []
+
+    for row in rows:
+        card_id = row["id"]
+        name = row["name"]
+        set_code = row["set_code"] or ""
+        collector_number = row["collector_number"] or ""
+
+        try:
+            data = _enrich_from_scryfall(name, set_code, collector_number)
+        except Exception as e:
+            errors.append(f"Error for '{name}': {e}")
+            skipped_count += 1
+            continue
+
+        if not data or "_error" in data:
+            err_msg = data.get("_error", f"Not found: '{name}'") if data else f"Not found: '{name}'"
+            errors.append(err_msg)
+            skipped_count += 1
+            continue
+
+        conn.execute(
+            """UPDATE collection_entries SET
+                type_line = ?, subtypes = ?, is_legendary = ?, is_basic = ?,
+                color_identity = ?, cmc = ?, mana_cost = ?,
+                oracle_text = ?, keywords = ?,
+                power = ?, toughness = ?,
+                rarity = ?, set_name = ?, edhrec_rank = ?,
+                tcg_price = ?, scryfall_id = ?, tcgplayer_id = ?,
+                category = ?,
+                updated_at = datetime('now')
+            WHERE id = ?""",
+            (
+                data.get("type_line", ""),
+                data.get("subtypes", "[]"),
+                data.get("is_legendary", 0),
+                data.get("is_basic", 0),
+                data.get("color_identity", "[]"),
+                data.get("cmc", 0.0),
+                data.get("mana_cost", ""),
+                data.get("oracle_text", ""),
+                data.get("keywords", "[]"),
+                data.get("power", ""),
+                data.get("toughness", ""),
+                data.get("rarity", ""),
+                data.get("set_name", ""),
+                data.get("edhrec_rank", 0),
+                data.get("tcg_price", 0.0),
+                data.get("scryfall_id", ""),
+                data.get("tcgplayer_id", ""),
+                data.get("category", "[]"),
+                card_id,
+            ),
+        )
+        conn.commit()
+        enriched_count += 1
+
+    return {
+        "total": total,
+        "enrichedCount": enriched_count,
+        "skippedCount": skipped_count,
+        "errors": errors[:50],  # cap error list
+    }
+
+
+# ══════════════════════════════════════════════════════════════
+# Auto-Classify Collection
+# ══════════════════════════════════════════════════════════════
+
+@router.post("/api/collection/auto-classify")
+async def auto_classify_collection():
+    """
+    Run auto-classification on all collection entries.
+    Uses oracle_text, type_line, and keywords to detect functional roles
+    (Ramp, Draw, Removal, Board Wipe, Anthem, Stax, etc.).
+    Only updates cards whose current category is empty or '[]'.
+    """
+    conn = _get_db_conn()
+    rows = conn.execute(
+        "SELECT id, oracle_text, type_line, keywords, category FROM collection_entries ORDER BY id"
+    ).fetchall()
+
+    total = len(rows)
+    classified_count = 0
+    skipped_count = 0
+
+    for row in rows:
+        card_id = row["id"]
+        existing_cats = row["category"] or "[]"
+
+        # Parse existing categories
+        try:
+            cats = json.loads(existing_cats)
+        except Exception:
+            cats = []
+
+        # Skip cards that already have manually-set categories
+        if cats:
+            skipped_count += 1
+            continue
+
+        oracle_text = row["oracle_text"] or ""
+        type_line = row["type_line"] or ""
+        keywords = row["keywords"] or "[]"
+
+        roles = _detect_card_roles(oracle_text, type_line, keywords)
+
+        if roles:
+            conn.execute(
+                "UPDATE collection_entries SET category = ?, updated_at = datetime('now') WHERE id = ?",
+                (json.dumps(roles), card_id),
+            )
+            classified_count += 1
+
+    conn.commit()
+
+    return {
+        "total": total,
+        "classifiedCount": classified_count,
+        "skippedCount": skipped_count,
+        "message": f"Classified {classified_count} cards, skipped {skipped_count} (already had categories)",
+    }
+
+
+@router.post("/api/collection/auto-classify-all")
+async def auto_classify_all_collection():
+    """
+    Force re-classify ALL collection entries, overwriting existing categories.
+    """
+    conn = _get_db_conn()
+    rows = conn.execute(
+        "SELECT id, oracle_text, type_line, keywords FROM collection_entries ORDER BY id"
+    ).fetchall()
+
+    total = len(rows)
+    classified_count = 0
+
+    for row in rows:
+        card_id = row["id"]
+        oracle_text = row["oracle_text"] or ""
+        type_line = row["type_line"] or ""
+        keywords = row["keywords"] or "[]"
+
+        roles = _detect_card_roles(oracle_text, type_line, keywords)
+        conn.execute(
+            "UPDATE collection_entries SET category = ?, updated_at = datetime('now') WHERE id = ?",
+            (json.dumps(roles), card_id),
+        )
+        classified_count += 1
+
+    conn.commit()
+
+    return {
+        "total": total,
+        "classifiedCount": classified_count,
+        "message": f"Re-classified all {classified_count} cards",
+    }
+
+
+# ══════════════════════════════════════════════════════════════
+# EDHREC Recommendations Cache + Endpoint
+# ══════════════════════════════════════════════════════════════
+
+@router.get("/api/collection/{cardId}/edhrec")
+async def get_card_edhrec(cardId: int):
+    """
+    Get EDHREC recommendations for a collection card.
+    For legendary creatures (potential commanders): fetch commander synergy data.
+    For other cards: fetch 'also played with' data.
+    """
+    conn = _get_db_conn()
+    row = conn.execute("SELECT * FROM collection_entries WHERE id = ?", (cardId,)).fetchone()
+    if not row:
+        raise HTTPException(404, f"Card with id {cardId} not found")
+
+    card = _row_to_dict(row)
+    name = card["name"]
+    type_line = card.get("type_line", "")
+    is_legendary = card.get("is_legendary", 0)
+
+    # Determine if it's a legendary creature (potential commander)
+    is_commander = bool(
+        is_legendary and
+        re.search(r"\bCreature\b", type_line, re.IGNORECASE)
+    )
+
+    slug = _to_edhrec_slug(name)
+    cache_key = f"edhrec:{'cmd' if is_commander else 'card'}:{slug}"
+
+    cached = _edhrec_cache_get(cache_key)
+    if cached:
+        return cached
+
+    recommendations = []
+    links = {
+        "edhrecPage": (
+            f"https://edhrec.com/commanders/{slug}"
+            if is_commander
+            else f"https://edhrec.com/cards/{slug}"
+        ),
+        "archidektSearch": f"https://archidekt.com/search/cards?q={name.replace(' ', '+')}",
+        "moxfieldSearch": f"https://www.moxfield.com/search?q={name.replace(' ', '+')}",
+    }
+
+    try:
+        if is_commander:
+            url = f"https://json.edhrec.com/pages/commanders/{slug}.json"
+        else:
+            url = f"https://json.edhrec.com/pages/cards/{slug}.json"
+
+        req = Request(url, headers=_API_HEADERS)
+        with urlopen(req, timeout=10) as resp:
+            data = json.loads(resp.read().decode("utf-8"))
+
+        if is_commander:
+            # Extract top cards from cardlists
+            container = data.get("container", {})
+            json_dict = container.get("json_dict", {})
+            for cardlist in json_dict.get("cardlists", []):
+                tag = cardlist.get("tag", "")
+                for cv in cardlist.get("cardviews", [])[:10]:
+                    rec_name = cv.get("name", "")
+                    if not rec_name:
+                        continue
+                    synergy = cv.get("synergy", 0.0)
+                    recommendations.append({
+                        "name": rec_name,
+                        "synergy": synergy,
+                        "role": tag or "Recommended",
+                    })
+        else:
+            # Extract "also played with" cards
+            container = data.get("container", {})
+            json_dict = container.get("json_dict", {})
+            for cardlist in json_dict.get("cardlists", []):
+                tag = cardlist.get("tag", "")
+                for cv in cardlist.get("cardviews", [])[:10]:
+                    rec_name = cv.get("name", "")
+                    if not rec_name:
+                        continue
+                    synergy = cv.get("synergy", 0.0)
+                    recommendations.append({
+                        "name": rec_name,
+                        "synergy": synergy,
+                        "role": tag or "Also Played With",
+                    })
+
+    except Exception as e:
+        log_deckgen.error(f"  Error fetching data for '{name}': {e}")
+        # Graceful degradation — return empty results
+
+    result = {
+        "recommendations": recommendations[:30],
+        "links": links,
+    }
+    _edhrec_cache_set(cache_key, result)
+    return result
+
+

routes/shared.py

@@ -0,0 +1,204 @@
+"""
+routes/shared.py
+================
+Single source of truth for everything every router needs.
+
+Contains:
+  - Config class + CFG singleton
+  - Logging setup + all named loggers
+  - FastAPI app instance
+  - In-memory state (active_batches, COMMANDER_META, PRECON_INDEX, AI_PROFILES)
+  - All Pydantic request/response models
+  - SQLite DB helpers (_get_db_conn, init_collection_db)
+  - Scryfall cache + enrichment helpers
+  - Collection helpers (_detect_card_roles, _classify_card_type,
+      _build_collection_filters, _parse_finish, _parse_csv_content,
+      _parse_text_line, _auto_infer_mapping, _row_to_dict, _add_image_url,
+      _snake_to_camel, VALID_SORT_FIELDS, _JSON_FIELDS)
+  - Deck helpers (_get_deck_or_404, _compute_deck_analysis, _check_ratio_limit,
+      _TYPE_TARGETS, _TYPE_PRIORITY, _classify_card_type)
+  - Import helpers (_import_from_url, _fetch_archidekt_deck,
+      _fetch_edhrec_average, _parse_text_decklist, _save_profile_to_dck,
+      _to_edhrec_slug, _http_get, _API_HEADERS, _edhrec_cache_get,
+      _edhrec_cache_set)
+  - Precon helpers (PRECON_DIR, PRECON_INDEX, download_precon_database,
+      load_precon_index, GITHUB_PRECON_URL, PRECON_CACHE_HOURS)
+  - Sim helpers (build_java_command, get_java17, BatchState, parse_dck_file,
+      AI_PROFILES, run_batch_subprocess, _run_process_blocking,
+      _run_deepseek_batch_thread, _load_deck_cards_by_name, _get_deepseek_brain)
+"""
+
+from __future__ import annotations
+
+import argparse
+import asyncio
+import csv
+import io
+import json
+import logging
+import logging.handlers
+import os
+import re
+import sqlite3
+import subprocess
+import sys
+import threading
+import time
+import uuid
+from collections import Counter
+from datetime import datetime
+from pathlib import Path
+from typing import Optional
+from urllib.request import urlopen, Request
+from urllib.error import URLError
+
+try:
+    from fastapi import FastAPI, HTTPException, BackgroundTasks, Request as FastAPIRequest
+    from fastapi.middleware.cors import CORSMiddleware
+    from fastapi.responses import FileResponse, JSONResponse, StreamingResponse
+    from fastapi.staticfiles import StaticFiles
+    from pydantic import BaseModel
+except ImportError:
+    print("ERROR: FastAPI not installed. Run: pip install fastapi uvicorn")
+    sys.exit(1)
+
+
+# ══════════════════════════════════════════════════════════════
+# Configuration
+# ══════════════════════════════════════════════════════════════
+
+# Configuration (imported from models/state)
+from models.state import Config, CFG
+
+
+# ══════════════════════════════════════════════════════════════
+# Logging Setup
+# ══════════════════════════════════════════════════════════════
+
+_LOG_FORMAT = "%(asctime)s %(levelname)-5s [%(name)s] %(message)s"
+_LOG_DATE_FMT = "%Y-%m-%d %H:%M:%S"
+_LOG_DIR = Path(os.environ.get("CAL_LOG_DIR", "logs"))
+_LOG_MAX_BYTES = int(os.environ.get("CAL_LOG_MAX_MB", "25")) * 1024 * 1024
+_LOG_BACKUP_COUNT = int(os.environ.get("CAL_LOG_BACKUP_COUNT", "5"))
+
+
+def setup_logging(level: int = logging.INFO) -> None:
+    _LOG_DIR.mkdir(parents=True, exist_ok=True)
+    root_logger = logging.getLogger("commander_ai_lab")
+    root_logger.setLevel(level)
+    if root_logger.handlers:
+        return
+    formatter = logging.Formatter(_LOG_FORMAT, datefmt=_LOG_DATE_FMT)
+    console = logging.StreamHandler(sys.stdout)
+    console.setLevel(level)
+    console.setFormatter(formatter)
+    root_logger.addHandler(console)
+    file_handler = logging.handlers.RotatingFileHandler(
+        _LOG_DIR / "commander-ai-lab.log",
+        maxBytes=_LOG_MAX_BYTES,
+        backupCount=_LOG_BACKUP_COUNT,
+        encoding="utf-8",
+    )
+    file_handler.setLevel(level)
+    file_handler.setFormatter(formatter)
+    root_logger.addHandler(file_handler)
+
+
+log          = logging.getLogger("commander_ai_lab.api")
+log_batch    = logging.getLogger("commander_ai_lab.batch")
+log_sim      = logging.getLogger("commander_ai_lab.sim")
+log_coach    = logging.getLogger("commander_ai_lab.coach")
+log_deckgen  = logging.getLogger("commander_ai_lab.deckgen")
+log_collect  = logging.getLogger("commander_ai_lab.collection")
+log_scan     = logging.getLogger("commander_ai_lab.scanner")
+log_ml       = logging.getLogger("commander_ai_lab.ml")
+log_cache    = logging.getLogger("commander_ai_lab.cache")
+log_pplx     = logging.getLogger("commander_ai_lab.pplx")
+
+
+# ══════════════════════════════════════════════════════════════
+# In-Memory State
+# ══════════════════════════════════════════════════════════════
+
+# In-Memory State (imported from models/state)
+from models.state import (
+    BatchState, active_batches, COMMANDER_META,
+    BUILTIN_COMMANDERS, load_commander_meta,
+)
+
+
+# Pydantic Models (imported from models/)
+from models.requests import (
+    StartRequest, ImportUrlRequest, ImportTextRequest, MetaFetchRequest,
+    CreateDeckRequest, UpdateDeckRequest, AddDeckCardRequest,
+    PatchDeckCardRequest, BulkAddRequest, BulkAddRecommendedRequest,
+    DeckGenerationSourceConfig, DeckGenerationRequest,
+    DeckGenV3Request, DeckGenV3SubstituteRequest,
+)
+from models.responses import (
+    StartResponse, StatusResponse, DeckInfo, GeneratedDeckCard,
+)
+
+
+# Collection Database (imported from services/database.py)
+from services.database import (
+    COLLECTION_DB_PATH, _get_db_conn, init_collection_db,
+    _JSON_FIELDS, VALID_SORT_FIELDS,
+    _row_to_dict, _snake_to_camel, _add_image_url, _build_collection_filters,
+)
+# Card Role Detection (imported from services/card_analysis.py)
+from services.card_analysis import _detect_card_roles
+
+
+# ══════════════════════════════════════════════════════════════
+# Deck Helpers (imported from services/deck_service.py)
+from services.deck_service import (
+    _TYPE_PRIORITY, _TYPE_TARGETS, _classify_card_type, _get_deck_or_404,
+    _compute_deck_analysis, _check_ratio_limit, _to_edhrec_slug,
+    parse_dck_file, _save_profile_to_dck, _load_deck_cards_by_name,
+)
+
+
+# ══════════════════════════════════════════════════════════════
+# Scryfall Cache (imported from services/scryfall.py)
+from services.scryfall import (
+    SCRYFALL_CACHE_DB_PATH, SCRYFALL_CACHE_TTL_SECONDS, _API_HEADERS,
+    _ScryfallCache, _scryfall_cache, _scryfall_lock, _scryfall_last_call,
+    _scryfall_rate_limit, _fetch_scryfall_api, _enrich_from_scryfall, _scryfall_fuzzy_lookup,
+)
+
+# ══════════════════════════════════════════════════════════════
+# Import Helpers (extracted to services/import_helpers.py)
+from services.import_helpers import (
+    _http_get, _fetch_archidekt_deck, _fetch_edhrec_average,
+    _import_from_url, _parse_text_decklist, _save_profile_to_dck,
+    _edhrec_cache, _EDHREC_CACHE_TTL, _edhrec_cache_get, _edhrec_cache_set,
+)
+
+# ══════════════════════════════════════════════════════════════
+# Collection Import Helpers (extracted to services/import_service.py)
+from services.import_service import (
+    _parse_finish, _parse_text_line, _auto_infer_mapping, _parse_csv_content,
+)
+
+# ══════════════════════════════════════════════════════════════
+# Precon Helpers (extracted to services/precon_service.py)
+from services.precon_service import (
+    PRECON_DIR, PRECON_INDEX, GITHUB_PRECON_URL, PRECON_CACHE_HOURS,
+    load_precon_index, _sanitize_filename, _deck_to_dck, download_precon_database,
+    _get_precon_dir,
+)
+
+# ══════════════════════════════════════════════════════════════
+# AI Profiles (imported from models/state)
+from models.state import AI_PROFILES
+
+# ══════════════════════════════════════════════════════════════
+# Java / Sim Helpers (extracted to services/forge_runner.py)
+from services.forge_runner import (
+    _find_java17, get_java17, build_java_command,
+    run_batch_subprocess, _run_process_blocking,
+    _run_deepseek_batch_thread, _get_deepseek_brain,
+)
+
+_ml_logging_enabled = True

scanner/pipeline.py

@@ -13,6 +13,10 @@ No local OCR or OpenCV needed — all recognition is done by the
 Ximilar cloud API using visual AI.
 """
 from typing import Callable, Optional
+import logging
+
+log = logging.getLogger("commander_ai_lab.scanner")
+
 
 from .ximilar_client import identify_cards, XimilarResult
 
@@ -160,7 +164,7 @@ def _ximilar_to_scan_result(xr: XimilarResult,
         try:
             scryfall_data = scryfall_lookup(xr.name)
         except Exception as e:
-            print(f"    [SCAN] Scryfall lookup failed for '{xr.name}': {e}")
+            log.error(f"    Scryfall lookup failed for '{xr.name}': {e}")
 
     if scryfall_data and scryfall_data.get("object") != "error":
         result.scryfall_id = scryfall_data.get("id", "")
@@ -181,6 +185,6 @@ def _ximilar_to_scan_result(xr: XimilarResult,
             # Extract image from scryfall URL pattern
             # e.g., https://scryfall.com/card/m11/149/lightning-bolt
             result.image_uri = ""  # Will be fetched by frontend via scryfall_id
-        print(f"    [SCAN] No Scryfall match for '{xr.name}' — using Ximilar data only")
+        log.info(f"    No Scryfall match for '{xr.name}' — using Ximilar data only")
 
     return result

scanner/ximilar_client.py

@@ -13,6 +13,10 @@ import json
 from typing import Optional, List
 from urllib.request import Request, urlopen
 from urllib.error import URLError, HTTPError
+import logging
+
+log = logging.getLogger("commander_ai_lab.scanner.ximilar")
+
 
 XIMILAR_ENDPOINT = "https://api.ximilar.com/collectibles/v2/tcg_id"
 
@@ -82,7 +86,7 @@ def identify_cards(image_bytes: bytes, api_key: str,
     )
 
     try:
-        print(f"    [XIMILAR] Sending {len(image_bytes):,} bytes to API (analyze_all={analyze_all})")
+        log.info(f"    Sending {len(image_bytes):,} bytes to API (analyze_all={analyze_all})")
         with urlopen(req, timeout=60) as resp:
             resp_data = json.loads(resp.read().decode("utf-8"))
     except HTTPError as e:
@@ -168,12 +172,12 @@ def _parse_response(resp_data: dict) -> List[XimilarResult]:
     if not results:
         results.append(XimilarResult(error="Ximilar could not identify any cards"))
 
-    print(f"    [XIMILAR] Identified {len(results)} card(s)")
+    log.info(f"    Identified {len(results)} card(s)")
     for r in results:
         if r.name:
-            print(f"    [XIMILAR]   - {r.name} ({r.set_code}) conf={r.confidence:.2f}")
+            log.info(f"    - {r.name} ({r.set_code}) conf={r.confidence:.2f}")
         elif r.error:
-            print(f"    [XIMILAR]   - ERROR: {r.error}")
+            log.error(f"    - ERROR: {r.error}")
 
     return results
 

scripts/bench_deckgen.py

@@ -0,0 +1,222 @@
+#!/usr/bin/env python3
+"""
+scripts/bench_deckgen.py
+========================
+Benchmark for _get_collection_for_colors() and _build_collection_summary().
+
+Seeds an in-memory SQLite DB with randomised collection_entries rows,
+then times the SQL-filtered query for various commander colour identities.
+
+Usage:
+    python scripts/bench_deckgen.py [--rows 1000]
+"""
+from __future__ import annotations
+
+import argparse
+import json
+import random
+import sqlite3
+import time
+from typing import List
+
+# ── Helpers ──────────────────────────────────────────────────────────────────
+
+ALL_COLORS = ["W", "U", "B", "R", "G"]
+
+CARD_TYPE_LINES = [
+    "Creature - Human Soldier",
+    "Creature - Elf Druid",
+    "Instant",
+    "Sorcery",
+    "Artifact",
+    "Enchantment",
+    "Land",
+    "Artifact Creature - Golem",
+    "Legendary Creature - Dragon",
+    "Planeswalker - Jace",
+]
+
+
+def _random_color_identity() -> list:
+    """Return a random colour identity (including ~15% chance of colorless)."""
+    if random.random() < 0.15:
+        return []
+    n = random.randint(1, 3)
+    return sorted(random.sample(ALL_COLORS, n))
+
+
+def _seed_db(conn: sqlite3.Connection, num_rows: int):
+    """Create collection_entries table and insert random rows."""
+    conn.execute("""
+        CREATE TABLE IF NOT EXISTS collection_entries (
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
+            name TEXT NOT NULL,
+            type_line TEXT DEFAULT '',
+            color_identity TEXT DEFAULT '[]',
+            cmc REAL DEFAULT 0,
+            mana_cost TEXT DEFAULT '',
+            oracle_text TEXT DEFAULT '',
+            keywords TEXT DEFAULT '[]',
+            quantity INTEGER DEFAULT 1,
+            edhrec_rank INTEGER DEFAULT 0,
+            scryfall_id TEXT DEFAULT '',
+            tcg_price REAL DEFAULT 0,
+            salt_score REAL DEFAULT 0,
+            is_game_changer INTEGER DEFAULT 0,
+            category TEXT DEFAULT ''
+        )
+    """)
+    conn.execute("""
+        CREATE INDEX IF NOT EXISTS idx_ce_quantity
+            ON collection_entries (quantity) WHERE quantity > 0
+    """)
+
+    rows = []
+    for i in range(num_rows):
+        ci = _random_color_identity()
+        qty = random.choice([0, 1, 1, 1, 2, 3])  # ~17% zero quantity
+        rows.append((
+            f"Card_{i:05d}",
+            random.choice(CARD_TYPE_LINES),
+            json.dumps(ci),
+            round(random.uniform(0, 8), 1),
+            "",
+            f"Sample oracle text for card {i}",
+            json.dumps([]),
+            qty,
+            random.randint(1, 20000),
+            f"scryfall-{i:05d}" if random.random() > 0.02 else "",
+            round(random.uniform(0, 50), 2),
+            round(random.uniform(0, 5), 2),
+            1 if random.random() < 0.05 else 0,
+            random.choice(["", "ramp", "draw", "removal", ""]),
+        ))
+
+    conn.executemany("""
+        INSERT INTO collection_entries
+            (name, type_line, color_identity, cmc, mana_cost, oracle_text,
+             keywords, quantity, edhrec_rank, scryfall_id, tcg_price,
+             salt_score, is_game_changer, category)
+        VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+    """, rows)
+    conn.commit()
+    print(f"  Seeded {num_rows} rows ({sum(1 for r in rows if r[7] > 0)} with qty > 0)")
+
+
+# ── Query under test (same as routes/deckgen.py) ────────────────────────────
+
+def get_collection_for_colors(conn: sqlite3.Connection, color_identity: list) -> list:
+    """Mirrors the new SQL-filtered implementation."""
+    allowed_json = json.dumps([c.upper() for c in color_identity]) if color_identity else '[]'
+
+    rows = conn.execute("""
+        SELECT scryfall_id, name, type_line, mana_cost, cmc,
+               oracle_text, keywords, quantity, edhrec_rank, color_identity
+        FROM collection_entries
+        WHERE quantity > 0
+          AND scryfall_id != ''
+          AND NOT EXISTS (
+              SELECT 1 FROM json_each(
+                  CASE WHEN collection_entries.color_identity IS NULL
+                            OR collection_entries.color_identity = ''
+                       THEN '[]'
+                       ELSE collection_entries.color_identity
+                  END
+              ) AS jc
+              WHERE jc.value NOT IN (
+                  SELECT value FROM json_each(?)
+              )
+          )
+    """, (allowed_json,)).fetchall()
+
+    valid = []
+    for row in rows:
+        card = dict(row)
+        ci = card.get("color_identity", "[]")
+        try:
+            card["color_identity"] = json.loads(ci) if isinstance(ci, str) else ci
+        except Exception:
+            card["color_identity"] = []
+        valid.append(card)
+    return valid
+
+
+def get_collection_for_colors_old(conn: sqlite3.Connection, color_identity: list) -> list:
+    """Old implementation: SELECT * + Python-side filter."""
+    rows = conn.execute(
+        "SELECT * FROM collection_entries WHERE quantity > 0"
+    ).fetchall()
+    valid = []
+    for row in rows:
+        card = dict(row)
+        card_ci = card.get("color_identity", "[]")
+        if isinstance(card_ci, str):
+            try:
+                card_ci = json.loads(card_ci)
+            except Exception:
+                card_ci = []
+        if color_identity and card_ci:
+            if not all(c in color_identity for c in card_ci):
+                continue
+        if not card.get("scryfall_id"):
+            continue
+        card["color_identity"] = card_ci
+        valid.append(card)
+    return valid
+
+
+# ── Main ─────────────────────────────────────────────────────────────────────
+
+COMMANDERS = [
+    ("Mono-W",    ["W"]),
+    ("WUB",       ["W", "U", "B"]),
+    ("WUBRG",     ["W", "U", "B", "R", "G"]),
+    ("Colorless", []),
+    ("Mono-G",    ["G"]),
+]
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--rows", type=int, default=1000)
+    args = parser.parse_args()
+
+    conn = sqlite3.connect(":memory:")
+    conn.row_factory = sqlite3.Row
+    _seed_db(conn, args.rows)
+
+    print(f"\n{'Commander':<14} {'Old ms':>8} {'New ms':>8} {'Old#':>6} {'New#':>6} {'Match':>6}")
+    print("-" * 60)
+
+    for label, ci in COMMANDERS:
+        # Old
+        t0 = time.perf_counter()
+        old_result = get_collection_for_colors_old(conn, ci)
+        old_ms = (time.perf_counter() - t0) * 1000
+
+        # New
+        t0 = time.perf_counter()
+        new_result = get_collection_for_colors(conn, ci)
+        new_ms = (time.perf_counter() - t0) * 1000
+
+        match = len(old_result) == len(new_result)
+        # NOTE: Colorless ([]) intentionally differs — old code had a bug that
+        # skipped the colour check entirely when color_identity was empty,
+        # returning ALL cards.  The new SQL correctly restricts to colourless.
+        if not ci:  # colorless commander
+            marker = "FIX" if len(new_result) < len(old_result) else "OK"
+        else:
+            marker = "OK" if match else "FAIL"
+        print(f"  {label:<12} {old_ms:>7.2f} {new_ms:>7.2f} {len(old_result):>6} {len(new_result):>6} {marker:>6}")
+
+        if not match and ci:  # only flag non-colorless mismatches
+            old_names = {c["name"] for c in old_result}
+            new_names = {c["name"] for c in new_result}
+            print(f"    Only in old: {len(old_names - new_names)} cards")
+            print(f"    Only in new: {len(new_names - old_names)} cards")
+
+    print()
+
+
+if __name__ == "__main__":
+    main()