fez5587/commander-ai-lab
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
128 Commits 31 Branches 3 Tags
8103a069926e1c4400784bcee0438bf7806b86d9
Commit Graph
100 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
KoalaTrapLord 8103a06992 Unity U1: scaffold unity-client/ folder + README (Epic #45) 2026-03-18 00:53:39 -04:00
KoalaTrapLord 4e8a999ebd feat: add Unity/Godot WebSocket server (Phase 1 stub) 
- pom.xml: add Java-WebSocket 1.5.6 dependency
- bridge/GameWsServer.java: WS server on ws://localhost:7654
- bridge/GameSession.java: Phase 1 stub game loop with human-input queue
- cli/LabCli.java: add ws-server subcommand via picocli
 2026-03-18 00:02:18 -04:00
KoalaTrapLord aa5e381c59 feat: add 4-player batch simulator React page with N-player API endpoint 
Adds SimulatorPage at /simulator with 4 deck slot selection, config panel, progress polling, per-seat win rate charts, and stats table. Backend: POST /api/sim/run-n-player endpoint using engine.run_n() for 2-4 player batch sims.
 2026-03-17 23:59:48 -04:00
KoalaTrapLord f4da7674b6 feat: React + TypeScript frontend migration (all 12 phases) 
Complete migration of the Commander AI Lab GUI from vanilla JS to React + TypeScript + Vite + Tailwind.

Phases implemented:
- Phase 1: Vite + React-TS scaffold with proxy config
- Phase 2: TypeScript type definitions (game.ts, constants.ts)
- Phase 3: useGameSocket hook + gameReducer
- Phase 4: Core components (PhaseBar, PlayerZone, CardChip, Board)
- Phase 5: NarrationPanel with auto-scroll
- Phase 6: MoveBar + HandCard
- Phase 7: GameOverOverlay
- Phase 8: App routing + GamePage assembly
- Phase 9: Lobby UI (LobbyPage, RoomPage, useLobbySock, spectator mode)
- Phase 10: Tailwind colour tokens
- Phase 11: Build integration + CI + README
- Phase 12: Test suite (26 tests, all passing)
 2026-03-17 22:58:54 -04:00
KoalaTrapLord 3f38a9885f Slim install script from 3.2MB to git clone approach (~8KB) - closes #31 
This script installs the Commander AI Lab environment by checking prerequisites, cloning the repository, and installing dependencies.
 2026-03-17 16:20:19 -04:00
KoalaTrapLord 940836db7e Add GitHub Actions CI workflow (lint + test on push) - closes #33 2026-03-17 16:18:50 -04:00
KoalaTrapLord fd74f157e8 Phase 4: Slim lab_api.py to thin main() - closes #25 
Refactor lab_api.py to streamline imports, improve logging, and enhance argument parsing.
 2026-03-17 16:17:53 -04:00
KoalaTrapLord 179a53c0af Implement Moxfield deck source adapter (closes #35) 2026-03-17 15:59:10 -04:00
KoalaTrapLord 554b166787 Phase 2d/2e/2f: Extract forge_runner, precon_service, import_service/helpers into services/ 2026-03-17 15:22:19 -04:00
KoalaTrapLord 874561f423 Phase 2c: Extract card analysis to services/card_analysis.py (#12) 2026-03-17 05:14:35 -04:00
KoalaTrapLord c7309d1ace Phase 2b: Extract deck service to services/deck_service.py (#11) 2026-03-17 05:05:52 -04:00
KoalaTrapLord 5aee93dfae Phase 1c: Extract datPhase 2a: Extract Scryfall client to services/scryfall.py (#10)abase service to services/database.py (#9) 2026-03-17 04:54:36 -04:00
KoalaTrapLord e14a6f9500 Phase 1c: Extract database service to services/database.py (#9) 2026-03-17 04:40:44 -04:00
KoalaTrapLord 7baae153ae Phase 1b: Extract Config, BatchState, AI_PROFILES into models/state.py (#8) 2026-03-17 01:57:22 -04:00
KoalaTrapLord eadf54c670 Phase 1a: Extract Pydantic models into models/requests.py + models/responses.py (#7) 2026-03-17 01:45:57 -04:00
KoalaTrapLord 312d13949d feat: extract ML endpoints into routes/ml.py (#36) 
Moved all 15 /api/ml/* endpoints, 4 state dicts, and 3 background
pipeline functions from lab_api.py into routes/ml.py. Wired via
app.include_router(ml_router) in lab_api.py.
 2026-03-17 00:41:31 -04:00
KoalaTrapLord 192c307ca7 refactor(#42): deduplicate sim thread functions in deepseek.py 2026-03-17 00:20:51 -04:00
KoalaTrapLord 5cb6e6ce81 refactor(#41): remove duplicate _get_deepseek_brain from deepseek.py, import from shared 2026-03-17 00:08:04 -04:00
KoalaTrapLord 01af4d7bcf cleanup(#38): delete stale ROUTES_WIRING.md 
This doc referenced the old routes._wire autoload pattern and only listed 5 routers. The project now uses 9 direct include_router() calls. Closes #38.
 2026-03-17 00:05:19 -04:00
KoalaTrapLord 301b9d38c8 fix(#40): use Path().parent.parent for correct src/ resolution in routes/deepseek.py 2026-03-17 00:02:22 -04:00
KoalaTrapLord 09b0708c4d fix(models): commander_tax int→dict to match engine per-commander tax tracking 2026-03-16 23:58:07 -04:00
KoalaTrapLord 1121e1cb26 feat: add commander_bonus weight to score_card() 
Commanders get +5.0 priority in AI spell selection, making the
AI prioritize casting its commander when affordable.
 2026-03-17 03:38:14 +00:00
KoalaTrapLord 4bade21220 feat: route commander deaths to command zone instead of graveyard 
Added _send_to_graveyard() helper that checks card.is_commander
and routes to command_zone (resetting tapped/turn_played) or
graveyard. Applied to all 4 death paths: removal, board wipe,
blocker death, and attacker death in combat.
 2026-03-17 03:36:36 +00:00
KoalaTrapLord 46421fe8c4 feat: add commander casting from command zone with tax tracking 
Second pass in _play_spells() checks command_zone for castable
commanders, applies commander_tax, taps lands, places on
battlefield, and increments tax by 2 for subsequent casts.
 2026-03-17 03:34:25 +00:00
KoalaTrapLord da5e9bd74f feat: set is_commander during deck loading, place in command_zone 
- Add commander_names param to run(), run_n(), _create_state()
- Commander card flagged is_commander=True and placed in
  player.command_zone instead of shuffled into library/hand
- Only first matching card is extracted (handles singleton rule)
- Backward compatible: commander_names defaults to None
 2026-03-17 03:31:35 +00:00
KoalaTrapLord 019a710acd feat: add command_zone and commander_tax to Player dataclass 
Adds command_zone (list[Card]) and commander_tax (int, +2 per cast)
to the base Player model so the headless engine can track commander
zone and tax natively without requiring CommanderPlayer wrappers.
 2026-03-17 03:18:06 +00:00
KoalaTrapLord 08509b753f fix: add httpx and websockets to runtime dependencies (fixes #37) 
Both were missing from [project.dependencies] in pyproject.toml and
requirements.txt. httpx is used in web/routers/art.py, and websockets
is needed for the FastAPI WebSocket endpoints in ws.py and lobby.py.

Closes #37
 2026-03-17 02:04:09 +00:00
Commander AI Lab 872f6c38ca security: remove hardcoded API keys, add .env support (fixes #26) 
- Remove plaintext Ximilar and Perplexity API keys from start-lab.bat
- Remove hardcoded key defaults from lab_api.py, lab_api_backup.py, lab_api_new.py
- start-lab.bat now reads from environment variables or .env file
- Add .env.example with placeholder values
- .env already in .gitignore (confirmed)
- Git history purged of all key values via git-filter-repo

Closes #26
 2026-03-17 02:00:15 +00:00
KoalaTrapLord 9f7fbef7f8 fix: implement all 19 simulator audit bug fixes 
Bug fixes across 7 Python files from the deep-dive audit:

engine.py (bugs 1-6):
- Fix summoning sickness (was disabled after turn 0)
- Move untap step before draw (correct MTG ordering)
- Rebuild playable list each iteration (fix mana overspend)
- Menace now allows 2+ blockers instead of preventing all
- Add double strike damage handling (blocked + unblocked)
- (Finish position sort already correct, no change needed)

game_state.py (bugs 7,9,18,19):
- Add .players property returning commander_players
- Add get_legal_moves()/apply_move() stubs
- Initialize commander zone in from_sim_state()
- Sync land_drop_used from engine stats

turn_manager.py (bugs 8,10,11):
- Use sim_state.get_battlefield() + card.tapped attr access
- Replace asyncio.get_event_loop() with get_running_loop()
- Fix APNAP iterations counter (always increment)
- Fix drawn.name and discarded.name (was dict .get())

threat_assessor.py (bugs 12,13,14):
- Use game_state.battlefield(seat) for card iteration
- Use Card dataclass attrs (c.type_line, c.oracle_text, etc)
- Move import re to module top level

models.py (bug 15):
- Word-boundary regex in has_keyword() (fixes reach/ward FPs)

prompt_builder.py (bug 16):
- Remove @staticmethod from module-level _category_label()

rules.py (bug 17):
- Evaluate ramp detection independently of is_removal flag
 2026-03-16 20:50:09 +00:00
KoalaTrapLord 2f117eedcb fix: Java 17 compat - inline ternary in this() call instead of local var 2026-03-16 20:34:05 +00:00
KoalaTrapLord 8f8abb79c6 fix: match actual Forge verbose log format - cast/played/Life:/Zone Change: patterns 2026-03-16 20:25:35 +00:00
KoalaTrapLord 78898f6da0 fix: regex patterns match Forge verbose output, single-winner resolution, 300s clock for 4p 2026-03-16 17:34:12 +00:00
KoalaTrapLord 69242a8f6e fix: remove duplicate numPlayers declaration in parseForgeOutput 2026-03-16 17:13:48 +00:00
KoalaTrapLord 2b6ccac3ea fix: hardcoded 3-element arrays in BatchRunner causing empty stats for 4-player games 2026-03-16 17:09:57 +00:00
KoalaTrapLord 94bb369866 feat: add --deck4/--commander4 support for 4-player games in LabCli + BatchRunner 2026-03-16 02:28:06 +00:00
KoalaTrapLord b57fc9dc2f fix: add missing stdlib imports (threading, os, re, csv, io, datetime) across route modules 2026-03-16 01:55:17 +00:00
KoalaTrapLord e1f2b5d89f fix: add missing os import to collection.py 2026-03-16 01:53:17 +00:00
KoalaTrapLord 9b354e019b fix: add missing Path import to collection.py 2026-03-16 01:42:37 +00:00
KoalaTrapLord ac3b76793b fix: add missing Path, os, threading imports to routes/collection.py 2026-03-15 21:05:35 -04:00
KoalaTrapLord c60719a0a9 refactor: extract deckgen (14) + coach (12) routes 
- routes/deckgen.py: classic generator, Perplexity V1, V3 generator + exports
  Helpers: _resolve_commander, _generate_deck, _build_collection_summary,
  _call_pplx_api, _fill_basic_lands, etc.
  Uses _coach._deck_gen_v3 via module reference for V3 state
- routes/coach.py: all coaching endpoints + init_coach_service(),
  _build_deck_report_from_db, globals (_coach_service, _deck_gen_v3, etc.)
- lab_api.py: 3265 -> 1063 lines (16 remaining: ML + SPA catch-all)
- Total: 90 extracted + 16 remaining = 106 endpoints verified
 2026-03-15 07:26:09 +00:00
KoalaTrapLord 87a2090c68 refactor: extract scanner (6) + deepseek/sim (9) routes 
- routes/scanner.py: scan, scan/add, re-enrich, auto-classify, auto-classify-all, edhrec
- routes/deepseek.py: sim/run, sim/status, sim/result, sim/run-from-deck,
  sim/run-deepseek, deepseek/connect, deepseek/status, deepseek/configure, deepseek/logs
- lab_api.py: 4464 -> 3265 lines (42 remaining endpoints)
- Total: 64 extracted + 42 remaining = 106 endpoints verified
 2026-03-15 06:56:02 +00:00
KoalaTrapLord 463b7c1483 Phase 8: Full multiplayer — lobby, private hands, LAN discovery, spectator mode, chat 2026-03-14 23:04:32 -04:00
KoalaTrapLord a093ab4949 Phase 7: Integration tests, AI benchmarking suite, conftest, pytest config, coverage setup 2026-03-14 21:13:26 -04:00
KoalaTrapLord b6b1616dad Phase 6: AI politics engine (deal-making, threat comms, spite logic) + targeting memory 2026-03-14 20:42:48 -04:00
KoalaTrapLord 92bb9b9445 Phase 5: FastAPI server, WebSocket game channel, Scryfall art proxy, REST API, static web client, tests 2026-03-14 20:36:06 -04:00
KoalaTrapLord 3c3dceb737 Phase 4: Add Pygame prototype GUI (gui/ package + tests/test_phase4.py) 2026-03-14 20:10:43 -04:00
KoalaTrapLord 605ff774ed Phase 3: Add turn_manager.py, threat_assessor.py, tests/test_phase3.py 2026-03-14 19:35:50 -04:00
KoalaTrapLord 441ba7dd74 Phase 2: Add ai_opponent.py, personality_prompts.py, tests/test_phase2.py 2026-03-14 16:53:43 -04:00
KoalaTrapLord d0e2824038 Phase 1: Add game_state.py, prompt_builder.py, state_logger.py 2026-03-14 16:43:16 -04:00
KoalaTrapLord 8a6fbfaa8a fix: 3 post-refactor bugs (COMMANDER_META, PRECON_INDEX, startup) 
Task 1: Move BUILTIN_COMMANDERS + load_commander_meta() to shared.py
        Fix import_routes.py to use _shared.COMMANDER_META (module ref)
Task 2: Rewrite precon.py to use _shared.PRECON_INDEX (module ref)
Task 3: Add @app.on_event('startup') safety net for uvicorn --reload
 2026-03-14 19:08:21 +00:00
KoalaTrapLord 36077c94ab refactor: extract 49 endpoints into 5 route modules 
- Extract collection (10), deckbuilder (18), precon (4), import (5), lab (12) routes
- All share helpers/models/state from routes/shared.py
- Gut lab_api.py from 8872 to 4487 lines (57 remaining endpoints)
- Delete routes/_wire.py and routes/_autoload.py (replaced by include_router)
- Add _scryfall_fuzzy_lookup to shared.py
- 106 total endpoints verified: 49 extracted + 57 remaining
 2026-03-14 18:55:32 +00:00
KoalaTrapLord 2c6748a736 feat(brain): swap default model to GPT-OSS 20B via Ollama (localhost:11434) 2026-03-14 14:34:14 -04:00
KoalaTrapLord 585cc1d4a3 feat(engine): load learned_weights.json automatically via load_weights() 2026-03-14 14:30:28 -04:00
KoalaTrapLord c52eb1c3ef refactor: extract all shared state, helpers, models into routes/shared.py 2026-03-14 03:15:12 -04:00
KoalaTrapLord d68b2368b4 fix: remove export_deck_to_sim and _edhrec_cache* from _wire.py (not yet in lab_api.py) 2026-03-14 03:02:45 -04:00
KoalaTrapLord 8d1a916f04 feat(ml): add update_weights.py — outcome-weighted RL learning from decision logs 2026-03-14 02:58:32 -04:00
KoalaTrapLord 2b2383f5c8 wire routes._autoload.register_startup(app) after CORSMiddleware 2026-03-14 02:02:53 -04:00
KoalaTrapLord 911fde9f5f wire routes._autoload.register_startup(app) after CORSMiddleware 2026-03-14 02:02:26 -04:00
KoalaTrapLord 6799f0c402 wire routes._autoload.register_startup(app) after CORSMiddleware 2026-03-14 01:57:18 -04:00
KoalaTrapLord da4e225db0 wire routes._autoload.register_startup(app) after CORSMiddleware 2026-03-14 01:55:50 -04:00
KoalaTrapLord 71428c89db docs: add ROUTES_WIRING.md — final integration instructions for lab_api.py 
Explains exactly what 2 lines to add to lab_api.py to activate
all 5 routers. Part of Batch 6/6.
 2026-03-14 01:50:47 -04:00
KoalaTrapLord c14402494b feat(routes): add _autoload.py — FastAPI startup event that wires routers 
Adds an @app.on_event('startup') hook that imports routes._wire.
This means lab_api.py needs zero changes — the routers self-register
when the app boots. Drop this file into the repo root and it works.

Alternatively, add to lab_api.py after app creation:
    from routes import _autoload  # noqa
 2026-03-14 01:49:52 -04:00
KoalaTrapLord 5ea512bfef feat(routes): add _wire.py — registers shared symbols and includes all routers 
Called via `import routes._wire` at the bottom of lab_api.py (after all
symbols are defined). Registers every shared object into routes.shared
and mounts all five APIRouters onto the FastAPI app.

Batch 6/6 — final wiring step.
 2026-03-14 01:49:18 -04:00
KoalaTrapLord d92345927f refactor: Batch 5 — routes/deckbuilder.py (deck CRUD, cards, analysis, recs, bulk-add, export) 2026-03-14 01:47:04 -04:00
KoalaTrapLord e1a46202b2 refactor: Batch 4 — routes/collection.py (collection CRUD, import, export, scryfall cache) 2026-03-14 01:46:33 -04:00
KoalaTrapLord 553e07980f refactor: Batch 3 — routes/import_routes.py and routes/precon.py 2026-03-14 01:45:58 -04:00
KoalaTrapLord b0666bb3fd refactor: Batch 2 — routes/lab.py (simulator, status, decks-list, profiles, analytics, deepseek) 2026-03-14 01:45:17 -04:00
KoalaTrapLord 4b1d2f71f5 refactor: Batch 1 — add routes/ package with shared state, config, and models 2026-03-14 01:44:37 -04:00
KoalaTrapLord 2d02fcc770 Add files via upload 2026-03-14 01:00:25 -04:00
KoalaTrapLord 3d00b68566 Delete lab_api_patched.py 2026-03-14 00:59:15 -04:00
KoalaTrapLord dda82ad1e6 Add files via upload 2026-03-14 00:55:54 -04:00
KoalaTrapLord 6c547664fb Delete lab_api.py 2026-03-14 00:55:32 -04:00
KoalaTrapLord d95f0b57c0 fix: commander card correctly flagged in deck tracker 
Two fixes:
1. _parse_text_decklist(): replace fragile startswith() section-header
   detection with case-insensitive regex so [Commander], [Main], [Deck]
   etc. are always recognised regardless of capitalisation.

2. add_deck_card endpoint: when a card is added with is_commander=1,
   also update the deck-level commander_name + commander_scryfall_id
   so the portrait and top-bar label stay in sync.
 2026-03-14 00:48:19 -04:00
KoalaTrapLord 264fb33957 fix: commander card correctly flagged in deck tracker 
Two fixes:
1. _parse_text_decklist(): replace fragile startswith() section header
   detection with proper case-insensitive regex so [Commander], [Main],
   [Deck] etc. are always recognised regardless of capitalisation.

2. add_deck_card endpoint: when a card is added with is_commander=1,
   also update the deck-level commander_name and commander_scryfall_id
   fields so the portrait and top-bar label stay in sync without
   requiring a full reload.
 2026-03-14 00:47:59 -04:00
KoalaTrapLord e4ebd84c04 test 2026-03-14 00:47:37 -04:00
KoalaTrapLord 5b072174f8 Add Phase 8 — Multiplayer (Human vs Human + AI) 2026-03-14 00:33:24 -04:00
KoalaTrapLord 661ca9618b Add GUI framework decision and two-phase frontend plan 2026-03-14 00:30:36 -04:00
KoalaTrapLord 7b5c9bdcc6 Update LLM model to GPT-OSS 20B (Q4_K_M) 2026-03-14 00:23:12 -04:00
KoalaTrapLord 6c5c55b637 Add Commander GUI + LLM Opponents roadmap 2026-03-14 00:20:41 -04:00
KoalaTrapLord f05067bbfe feat(brain): pipe GameResult winner_seat + per-player life totals into flush_log() for RL training 2026-03-14 00:13:50 -04:00
KoalaTrapLord dfc1f46097 fix(ui): merge duplicate .stat-card declarations in premium layer 2026-03-14 00:10:14 -04:00
KoalaTrapLord 6cafa36ce4 feat(rules): add load_weights() / save_weights() for simulation-driven weight learning 2026-03-13 23:58:59 -04:00
KoalaTrapLord 7df6b95b4c feat(ui): add premium upgrade layer to global.css 2026-03-13 23:51:47 -04:00
KoalaTrapLord 812e5b34ca feat(tests): add pytest infrastructure and test coverage for models, rules, and sim engine (closes #28) 2026-03-13 22:30:40 -04:00
KoalaTrapLord 8c89119ba1 chore: add requirements.txt for pip install compatibility (closes #29) 2026-03-13 20:53:31 -04:00
KoalaTrapLord b1891c58f8 fix(deps): declare all runtime dependencies in pyproject.toml (closes #29) 2026-03-13 20:53:18 -04:00
KoalaTrapLord b5e7705100 feat(ui): 4-player deck slots, seat colors, chart support (Task 2C) 2026-03-13 19:08:53 -04:00
KoalaTrapLord 4c7c3495fb feat(api): accept 2-4 decks in batch start, dynamic Java CLI args (Task 2C) 2026-03-13 18:59:25 -04:00
KoalaTrapLord c1371849e5 feat(api): accept 2-4 decks in batch start, dynamic Java CLI args (Task 2C) 2026-03-13 18:58:33 -04:00
KoalaTrapLord 3190fa3702 feat(api): accept 2-4 decks in batch start, dynamic Java CLI args (Task 2C) 2026-03-13 18:56:58 -04:00
KoalaTrapLord 9f48e056ff feat(ui): 4-player deck grid, seat colors, responsive breakpoints (Task 2C) 2026-03-13 18:55:17 -04:00
KoalaTrapLord 841e861bea feat: N-player deepseek engine (Task 2A — Issue #27) 
- run_n() accepts 2-4 decks; run() wraps for backward compat
- _select_attack_target() heuristic for multiplayer combat targeting
- N-player elimination order tracking with finish positions
- Removal scans all opponents for scariest creature
- ML snapshot captures all N players
- Log entries use per-seat boards dict with legacy boardA/boardB for 2-player
 2026-03-13 16:58:23 -04:00
KoalaTrapLord 0be8841cec feat: N-player engine turn loop (Task 2A — Issue #27) 
- engine.py: new run_n() method accepts 2-4 decks; run() wraps for backward compat
- _select_attack_target() heuristic: attack weakest non-eliminated opponent
- N-player elimination order tracking with correct finish positions
- Removal spells scan all opponents for best target
- Game logs use per-seat boards dict with boardA/boardB legacy keys for 2-player
- Tested: 2/3/4-player, elimination order, attack targeting
 2026-03-13 16:54:55 -04:00
KoalaTrapLord ddc0e4fc05 fix: preserve is_commander flag when loading .dck files 
Two places lost the commander flag:

1. _load_deck_cards_by_name() .dck fallback — used a flat in_cards
   boolean for [Main], [Deck], and [Commander] sections so it couldn't
   tell which section a card came from. Now tracks current_section and
   sets is_commander=1 for cards in [Commander].

2. parse_dck_file() — stored section='Commander' in the card dict but
   the frontend filters on c.is_commander, not c.section. Now includes
   is_commander: 1/0 alongside the existing section field.
 2026-03-13 20:20:18 +00:00
KoalaTrapLord a1fd290d5f refactor: adopt Python logging module with named loggers 
Replace all print() calls with structured logging using the Python
logging module. This provides:

- Named loggers per subsystem: commander_ai_lab.api, .batch, .sim,
  .coach, .deckgen, .collection, .scanner, .ml, .cache, .pplx
- Dual output: console (StreamHandler) + rotating file (RotatingFileHandler)
  at logs/commander-ai-lab.log (25 MB x 5 backups = 125 MB max)
- Configurable via env vars: CAL_LOG_DIR, CAL_LOG_MAX_MB, CAL_LOG_BACKUP_COUNT
- --verbose / -v flag for DEBUG-level output
- overnight-run.py: separate rotating log at logs/overnight-run.log (10 MB x 5)
- Sub-module loggers in ml/, scanner/, deck_sources/, src/ all propagate
  to the root commander_ai_lab logger
- Stripped redundant [TAG] prefixes from messages (logger name carries context)
- Format: '2026-03-13 14:00:00 INFO  [commander_ai_lab.batch] message'

Files changed: lab_api.py, overnight-run.py, and 11 sub-module .py files
176 total print() calls converted (101 in lab_api.py, 76 in sub-modules)
 2026-03-13 18:09:52 +00:00
KoalaTrapLord 8d2e6199db feat: Scryfall response caching with SQLite 
- _ScryfallCache class: thread-safe SQLite cache for raw Scryfall JSON
  responses, keyed by name|set_code|collector_number, 7-day TTL
- _fetch_scryfall_api() wrapper: checks cache first, live HTTP on miss,
  caches successful responses + resolves spelling normalization
- Refactored _enrich_from_scryfall() to use cache — all 3 callers
  (collection import, deck card add, re-enrich) benefit automatically
- Rate limiting only on actual HTTP calls, not cache hits
- REST endpoints: GET/DELETE /api/cache/scryfall, POST .../evict
- Fixed VACUUM-inside-transaction bug in clear()
- Configurable TTL via SCRYFALL_CACHE_TTL env var
 2026-03-13 15:25:30 +00:00
KoalaTrapLord 3d8e81f88e Phase 2: score_card() num_opponents scaling + enrich_card() determinism fix (#32) 
- score_card() now accepts num_opponents: int parameter (default 1)
- Keywords/effects scale with pod size via _MULTI_OPP_SCALING dict
- Multiplayer-specific oracle patterns: 'each opponent', pillowfort, 'each player'
- Board wipe and counter spell bonuses scale with opponent count
- Expanded keyword coverage (double strike, first strike, hexproof, indestructible,
  menace, vigilance, ward, infect, annihilator, cascade, defender)
- Triggered ability bonuses: ETB, attack triggers, dies triggers
- enrich_card() replaced random.randint() with hashlib.md5 name-based hash
  for deterministic CMC estimation across games (#32)
- Backward compatible: existing score_card(card, weights) calls unaffected
 2026-03-12 20:22:44 +00:00
KoalaTrapLord 648e1b8d38 Fix simulator button stuck on Running — to_dict() backward compat 
GameResult.to_dict() now returns both legacy keys (winner, playerA, playerB)
that the UI expects AND new N-player keys (winningSeat, playerResults) for
the Java BatchResult schema.
 2026-03-12 18:25:22 +00:00
KoalaTrapLord de4f119d40 Phase 1: N-player data model refactoring 
- New PlayerResult dataclass with seat_index, name, life, eliminated, finish_position, stats
- Refactored GameResult to use players: list[PlayerResult] + winner_seat (no hardcoded 2-player fields)
- Backward-compat @property accessors (player_a_stats, player_b_stats, winner, etc.) for existing consumers
- SimState.battlefields is now per-player (list-of-lists) with helper methods:
  get_battlefield(), add_to_battlefield(), remove_from_battlefield(), all_battlefield_cards()
- Updated engine.py, deepseek_engine.py, deepseek_brain.py to use per-player battlefields
- GameResult.to_dict() matches Java BatchResult schema (playerResults array)
- lab_api.py and experiments.py unchanged — backward-compat properties handle all existing access patterns
 2026-03-12 18:10:57 +00:00
KoalaTrapLord 731b40b15f Fix duplicate deck cards from collection JOIN fan-out 
When collection_entries has multiple rows with the same scryfall_id
(e.g. Counterspell imported from two different sets), the LEFT JOIN
on deck_cards produced duplicate rows — making it appear like 2 copies
were added or removed.

Fixed all 6 deck_cards-to-collection_entries JOINs to use a
GROUP BY scryfall_id subquery, guaranteeing 1:1 cardinality.
 2026-03-12 17:51:20 +00:00