Backend:
- Add get_teams_by_owner() to SBA API client
- Update /api/auth/me to return user's teams from SBA API
- Add sba_current_season config setting (default: 13)
Frontend:
- Replace hardcoded myTeamId with computed from auth store teams
- Fix isMyTurn logic to check actual team ownership
- Update REFACTORING_PLAN.json
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Frontend UX improvements:
- Single-click Discord OAuth from home page (no intermediate /auth page)
- Auto-redirect authenticated users from home to /games
- Fixed Nuxt layout system - app.vue now wraps NuxtPage with NuxtLayout
- Games page now has proper card container with shadow/border styling
- Layout header includes working logout with API cookie clearing
Games list enhancements:
- Display team names (lname) instead of just team IDs
- Show current score for each team
- Show inning indicator (Top/Bot X) for active games
- Responsive header with wrapped buttons on mobile
Backend improvements:
- Added team caching to SbaApiClient (1-hour TTL)
- Enhanced GameListItem with team names, scores, inning data
- Games endpoint now enriches response with SBA API team data
Docker optimizations:
- Optimized Dockerfile using --chown flag on COPY (faster than chown -R)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Create frontend-sba/.env.example and frontend-pd/.env.example templates
- Fix hardcoded allowedHosts in nuxt.config.ts (now reads NUXT_ALLOWED_HOSTS)
- Add NUXT_ALLOWED_HOSTS support to frontend-pd/nuxt.config.ts
- Update docker-compose.yml with missing env vars:
- FRONTEND_URL, DISCORD_SERVER_REDIRECT_URI
- ALLOWED_DISCORD_IDS, WS_HEARTBEAT_INTERVAL, WS_CONNECTION_TIMEOUT
- NUXT_ALLOWED_HOSTS for both frontends
- Create docker-compose.prod.yml for production overrides
- Update root .env.example with new variables
- Add "Multi-Domain Deployment" section to README.md with checklist
- Update all CLAUDE.md files with environment configuration docs
- Remove obsolete 'version' attribute from docker-compose files
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
The outs_before field in play records was incorrectly storing the
outs count AFTER _apply_play_result() modified state.outs, rather
than the value before the play resolved.
Fix: Capture state.outs before calling _apply_play_result() and
pass it explicitly to _save_play_to_db() as a required parameter.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
When recovering game state after reconnect/refresh, the old code:
1. Only looked at the last play overall (regardless of team)
2. Used that play's batting order for the currently batting team
3. Reset the non-batting team's index to 0
This caused the wrong batter to appear after inning changes - e.g.,
batter #1 showing instead of #5 in top of 2nd after bottom of 1st ended.
Fix: Now recovers each team's batter index separately by filtering
plays by half ("top" = away team, "bottom" = home team) and finding
each team's last at-bat independently.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
After changing runners_advanced from list[tuple] to list[RunnerAdvancementData],
three locations were still trying to unpack as tuples causing "cannot unpack
non-iterable RunnerAdvancementData object" error on triples/doubles with runners.
Fixed locations:
- app/core/game_engine.py:772 - _apply_play_result() advancement map
- app/core/game_engine.py:1081 - _save_play() finals lookup
- terminal_client/display.py:150 - Runner movement display
Also renamed refactor_overview.md to game-engine-play-tracking-design.md
for clarity and updated the example code to use dataclass access pattern.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Root cause: WebSocket handler had stale state reference after
resolve_manual_play() completed. The handler's state object still had
the old batter index, and calling state_manager.update_state()
overwrote the game engine's updated state.
Fix: Re-fetch state from state_manager AFTER resolve_manual_play()
returns, ensuring we get the state with the advanced batter index.
Also improved logging in _prepare_next_play() to show batting_order
for easier debugging.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Root cause: Auth middleware was commented out on game detail page
([id].vue), causing SSR to render without checking authentication.
Safari's client-side auth check wasn't reaching the backend due to
caching behavior, resulting in "Auth: Failed" display.
Changes:
- Re-enabled middleware: ['auth'] in pages/games/[id].vue
- Added /api/auth/ws-token endpoint for Safari WebSocket fallback
- Added expires_minutes param to create_token() for short-lived tokens
- Added token query param support to WebSocket handlers
- Updated SAFARI_WEBSOCKET_ISSUE.md documentation
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Backend changes:
- Modified request_game_state handler to fetch plays from database
- Convert Play DB models to frontend-compatible PlayResult dicts
- Emit game_state_sync event with state + recent_plays array
Frontend changes:
- Added deduplication by play_number in addPlayToHistory()
- Prevents duplicate plays when game_state_sync is received
Field mapping from Play model:
- hit_type -> outcome
- result_description -> description
- batter_id -> batter_lineup_id
- batter_final -> batter_result
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
New Tests:
- test_exceptions.py: Comprehensive tests for custom GameEngineError
hierarchy including AuthorizationError, DatabaseError, etc.
Scripts:
- stop-services.sh: Enhanced to kill entire process trees and
clean up orphan processes
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Game Engine Improvements:
- State manager: Add game eviction with configurable max games, idle
timeout, and memory limits
- Game engine: Add resource cleanup on game completion
- Play resolver: Enhanced RunnerAdvancementData with lineup_id for
player name resolution in play-by-play
- Substitution manager: Minor improvements
Test Coverage:
- New test_game_eviction.py with 13 tests for eviction scenarios
- Updated state_manager tests for new functionality
- Updated play_resolver tests for lineup_id handling
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Add rate_limit.py middleware with per-client throttling and cleanup task
- Add pool_monitor.py for database connection pool health monitoring
- Add custom exceptions module (GameEngineError, DatabaseError, etc.)
- Add config settings for eviction intervals, session timeouts, memory limits
- Add unit tests for rate limiting and pool monitoring
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Changed cookie SameSite policy from Lax to None with Secure=true
for Safari ITP compatibility
- Fixed Nuxt composable context issue: moved useRuntimeConfig() from
connect() callback to composable setup phase (required in Nuxt 4)
- Added GET /logout endpoint for easy browser-based logout
- Improved loading overlay with clear status indicators and action
buttons (Retry, Re-Login, Dismiss)
- Added error handling with try-catch in WebSocket connect()
- Documented issue and fixes in .claude/SAFARI_WEBSOCKET_ISSUE.md
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Fixed missing PlayOutcome.HIT_BY_PITCH handling in resolve_outcome().
HBP was being submitted via WebSocket but rejected with error:
"Unhandled outcome: PlayOutcome.HIT_BY_PITCH"
Changes:
- Added HBP case to play_resolver.py (lines 430-446)
- HBP uses same forced runner advancement logic as WALK
- HBP is NOT classified as is_walk=True (correct for stats)
- Added 2 unit tests: test_hit_by_pitch, test_hit_by_pitch_bases_loaded
Test results: 981/981 passing (100%)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Added rollback_play WebSocket handler (handlers.py:1632)
- Accepts game_id and num_plays (default: 1)
- Validates game state and play count
- Broadcasts play_rolled_back and game_state_update events
- Full error handling with rate limiting
- Added undoLastPlay action to useGameActions composable
- Emits rollback_play event to backend
- Added Undo button to game page ([id].vue)
- Amber floating action button with undo arrow icon
- Positioned above substitutions button
- Only visible when game is active and has plays
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Resolved WebSocket connection issues and games list loading on iPad:
- cookies.py: Added is_secure_context() to set Secure flag when accessed
via HTTPS even in development mode (Safari requires this)
- useWebSocket.ts: Changed auto-connect from immediate watcher to
onMounted hook for safer SSR hydration
- games/index.vue: Replaced onMounted + fetchGames() with useAsyncData
for SSR data fetching with proper cookie forwarding
- Updated COOKIE_AUTH_IMPLEMENTATION.md with new issues and solutions
- Updated composables/CLAUDE.md with auto-connect pattern documentation
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Two related bug fixes for gameplay accuracy:
**Backend - play_resolver.py**:
- Fixed DOUBLE2 advancement: Runners now advance exactly 2 bases
* 1st → 3rd, 2nd → home, 3rd → home
* Was incorrectly advancing all runners to home
- Fixed DOUBLE3 advancement: Runners now advance exactly 3 bases
* All runners score (1st+3=4, 2nd+3=5→4, 3rd+3=6→4)
* Updated docstrings for clarity
**Frontend - ManualOutcomeEntry.vue**:
- Fixed hit location requirement logic
* Now requires hit location when runners on base (any outs)
* Was incorrectly restricting to only when outs < 2
* Hit location determines runner advancement regardless of outs
These fixes ensure accurate Strat-O-Matic gameplay simulation.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Backend enhancements for real-time decision workflow:
**New Features**:
- decision_required event emission when game starts and after each decision
- Quick-create endpoint (/games/quick-create) for rapid testing with pre-configured lineups
- WebSocket connection manager integration in GameEngine
**Changes**:
- game_engine.py: Added _emit_decision_required() method and set_connection_manager()
- game_engine.py: Emit decision_required on game start with 5-minute timeout
- games.py: New /quick-create endpoint with Team 35 vs Team 38 lineups
- main.py: Wire connection manager to game_engine singleton
- state_manager.py: Enhanced state management for decision phases
- play_resolver.py: Improved play resolution logic
- handlers.py: Updated WebSocket handlers for new workflow
- backend/CLAUDE.md: Added WebSocket protocol spec reference
**Why**:
Eliminates polling - frontend now gets real-time notification when decisions are needed.
Quick-create saves 2 minutes of lineup setup during each test iteration.
**Testing**:
- Manual testing with terminal client
- WebSocket event flow verified with live frontend
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
This commit fixes two critical bugs in the game engine and updates tests
to match current webhook behavior:
1. State Recovery - Batter Advancement (operations.py:545-546)
- Added missing batting_order and outs_recorded fields to plays dictionary
- These fields exist in database but weren't loaded by load_game_state()
- Root cause: Batter index was correctly recovered but current_batter remained
at placeholder (batting_order=1) because recovery logic couldn't find the
actual batting_order from last play
- Fix enables proper batter advancement after backend restarts
2. Flyball Descriptions - 2 Outs Logic (runner_advancement.py)
- Made flyball descriptions dynamic based on outs and actual base runners
- FLYOUT_A (Deep): Lines 1352-1363
- FLYOUT_B (Medium): Lines 1438-1449
- FLYOUT_BQ (Medium-shallow): Lines 1521-1530
- With 2 outs: "3rd out, inning over" (no advancement possible)
- With 0-1 outs: Dynamic based on runners ("R3 scores" only if R3 exists)
- Game logic was already correct (runs_scored=0), only descriptions were wrong
- Fixed method signatures to include hit_location parameter for all flyball methods
- Updated X-check function calls to pass hit_location parameter
3. Test Updates
- Fixed test expectation in test_flyball_advancement.py to match corrected behavior
- Descriptions now only show runners that actually exist (no phantom "R2 DECIDE")
- Auto-fixed import ordering with Ruff
- Updated websocket tests to match current webhook behavior:
* test_submit_manual_outcome_success now expects 2 broadcasts (play_resolved + game_state_update)
* test_submit_manual_outcome_missing_required_location updated to reflect hit_location now optional
Testing:
- All 739 unit tests passing (100%)
- Verified batter advances correctly after state recovery
- Verified flyball with 2 outs shows correct description
- Verified dynamic descriptions only mention actual runners on base
- Verified websocket handler broadcasts work correctly
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Remove accidentally created backend/pages directory (Vue files belong in frontend).
Added pages/ to backend .gitignore to prevent future confusion.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Complete implementation of pre-game setup flow allowing players to create games
and submit lineups before gameplay starts.
Backend Changes:
- Extended games.py with create game, lineup submission, and game start endpoints
- Added teams.py roster endpoint with season filtering
- Enhanced SBA API client with player data fetching and caching
- Comprehensive validation for lineup submission (position conflicts, DH rules)
Frontend Changes:
- Redesigned create.vue with improved team selection and game options
- Enhanced index.vue with active/pending game filtering and navigation
- Added lineup/[id].vue for interactive lineup builder with drag-and-drop
- Implemented auth.client.ts plugin for client-side auth initialization
- Added comprehensive TypeScript types for API contracts
- Updated middleware for better auth handling
Key Features:
- Game creation with home/away team selection
- Full lineup builder with position assignment and batting order
- DH rule validation (pitcher can be excluded from batting order)
- Season-based roster filtering (Season 3)
- Auto-start game when both lineups submitted
- Real-time game list updates
Workflow:
1. Create game → select teams → set options
2. Submit home lineup → validate positions/order
3. Submit away lineup → validate positions/order
4. Game auto-starts → navigates to game page
5. WebSocket connection → loads game state
Ready for Phase F4 - connecting gameplay UI to complete the at-bat loop.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
## Authentication Implementation
### Backend
- Implemented complete Discord OAuth flow in auth.py:
* POST /api/auth/discord/callback - Exchange code for tokens
* POST /api/auth/refresh - Refresh JWT tokens
* GET /api/auth/me - Get authenticated user info
* GET /api/auth/verify - Verify auth status
- JWT token creation with 7-day expiration
- Refresh token support for session persistence
- Bearer token authentication for Discord API calls
### Frontend
- Created auth/login.vue - Discord OAuth initiation page
- Created auth/callback.vue - OAuth callback handler with states
- Integrated with existing auth store (already implemented)
- LocalStorage persistence for tokens and user data
- Full error handling and loading states
### Configuration
- Updated backend .env with Discord OAuth credentials
- Updated frontend .env with Discord Client ID
- Fixed redirect URI to port 3001
## SBA API Integration
### Backend
- Extended SbaApiClient with get_teams(season, active_only=True)
- Added bearer token auth support (_get_headers method)
- Created /api/teams route with TeamResponse model
- Registered teams router in main.py
- Filters out IL (Injured List) teams automatically
- Returns team data: id, abbrev, names, color, gmid, division
### Integration
- Connected to production SBA API: https://api.sba.manticorum.com
- Bearer token authentication working
- Successfully fetches ~16 active Season 3 teams
## Documentation
- Created SESSION_NOTES.md - Current session accomplishments
- Created NEXT_SESSION.md - Game creation implementation guide
- Updated implementation/NEXT_SESSION.md
## Testing
- ✅ Discord OAuth flow tested end-to-end
- ✅ User authentication and session persistence verified
- ✅ Teams API returns real data from production
- ✅ All services running and communicating
## What Works Now
- User can sign in with Discord
- Sessions persist across reloads
- Backend fetches real teams from SBA API
- Ready for game creation implementation
## Next Steps
See .claude/NEXT_SESSION.md for detailed game creation implementation plan.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Reviewed all remaining high-severity issues from code review:
Issues #6, #7 (Input validation): Already implemented in validators.py
- hold_runners validation: lines 71-77
- steal_attempts validation: lines 156-165
- Called from submit_defensive_decision and submit_offensive_decision
Issue #8 (Hardcoded inning limit): Deferred to next sprint
- Requires config system changes across validators.py and game_models.py
- Appropriate for technical debt phase
Issue #9 (Cleanup on abandon): Already fixed (part of Issue #3)
- _cleanup_game_resources() called in end_game() at line 1109
Issue #10 (Direct state mutation): Architectural acknowledgment
- Current pattern with debug logging throughout
- Consider immutable state pattern for v2 if auditability needed
Issue #11 (Logger singleton): Verified correct
- Module-level singleton at line 34, outside class
All high-priority issues now addressed or appropriately deferred.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Issue #5 from code review - resolve_play and resolve_manual_play shared
~70% of their code causing maintenance burden and divergent behavior risk.
Changes:
- Extracted common finalization logic to new _finalize_play() method
- resolve_play reduced from ~150 lines to ~60 lines
- resolve_manual_play reduced from ~135 lines to ~60 lines
- Single source of truth for: roll tracking, state capture, transaction
handling, inning advance, cleanup, and state updates
Benefits:
- Changes to play finalization only need to be made in one place
- Reduced risk of divergent behavior between resolution modes
- Better testability and maintainability
All 739 unit tests passing (1 flaky probabilistic test occasionally fails).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Issue #4 from code review - multi-step database operations were not
wrapped in transactions, risking partial state on failure.
Changes:
- Modified save_play() and update_game_state() in DatabaseOperations
to accept optional session parameter for transaction grouping
- Wrapped resolve_play() DB operations in single atomic transaction
- Wrapped resolve_manual_play() DB operations in single atomic transaction
- Transaction commits atomically or rolls back entirely on failure
Pattern: When session provided, use flush() for IDs without committing;
caller controls transaction. When no session, create internal transaction.
All 739 unit tests passing.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- test_game_data.py: Static lineup data based on real SBA Game 2519
- West Virginia Black Bears vs Columbus Hydra
- Complete lineup data with player info
- create_test_game.py: Script to end active games and create fresh test game
- Usage: uv run python scripts/create_test_game.py
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Backend fixes:
- state_manager: Properly recover current_pitcher and current_catcher from
fielding team during game state recovery (fixes pitcher badge not showing)
- handlers: Add headshot field to lineup data, use lineup_service for proper
player data loading on cache miss
- lineup_service: Minor adjustments for headshot support
Frontend fixes:
- player.ts: Update Lineup type to match WebSocket event format
- lineup_id (was 'id'), card_id fields
- player.headshot for UI circles
- Optional fields for event variations
- CurrentSituation.vue: Adapt to updated type structure
- Substitution selectors: Use updated Lineup type fields
This fixes the issue where pitcher badge wouldn't show after game recovery
because current_pitcher was being set from batting team instead of fielding team.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Created backend/.claude/DATABASE_SCHEMA.md with full table details
- Updated database/CLAUDE.md to reference the new schema doc
- Preserves valuable reference material while keeping CLAUDE.md concise
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Created centralized services for SBA player data fetching at lineup creation:
Backend - New Services:
- app/services/sba_api_client.py: REST client for SBA API (api.sba.manticorum.com)
with batch player fetching and caching support
- app/services/lineup_service.py: High-level service combining DB operations
with API calls for complete lineup entries with player data
Backend - Refactored Components:
- app/core/game_engine.py: Replaced raw API calls with LineupService,
reduced _prepare_next_play() from ~50 lines to ~15 lines
- app/core/substitution_manager.py: Updated pinch_hit(), defensive_replace(),
change_pitcher() to use lineup_service.get_sba_player_data()
- app/models/game_models.py: Added player_name/player_image to LineupPlayerState
- app/services/__init__.py: Exported new LineupService components
- app/websocket/handlers.py: Enhanced lineup state handling
Frontend - SBA League:
- components/Game/CurrentSituation.vue: Restored player images with fallback
badges (P/B letters) for both mobile and desktop layouts
- components/Game/GameBoard.vue: Enhanced game board visualization
- composables/useGameActions.ts: Updated game action handling
- composables/useWebSocket.ts: Improved WebSocket state management
- pages/games/[id].vue: Enhanced game page with better state handling
- types/game.ts: Updated type definitions
- types/websocket.ts: Added WebSocket type support
Architecture Improvement:
All SBA player data fetching now goes through LineupService:
- Lineup creation: add_sba_player_to_lineup()
- Lineup loading: load_team_lineup_with_player_data()
- Substitutions: get_sba_player_data()
All 739 unit tests pass.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Removed the incorrect restriction that squeeze bunt cannot be used with bases
loaded. The only requirements for squeeze bunt are:
- Runner on third base
- Not with 2 outs
Updated validator and test to reflect correct rule.
Test results: 739/739 passing (100%)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Removed the unused alignment field from DefensiveDecision model and all
related code across backend and frontend.
Backend changes:
- models/game_models.py: Removed alignment field and validator
- terminal_client/display.py: Removed alignment from display
- core/ai_opponent.py: Updated log message
- tests/unit/models/test_game_models.py: Removed alignment tests
- tests/unit/core/test_validators.py: Removed alignment validation test
Frontend changes:
- types/game.ts: Removed alignment from DefensiveDecision interface
- components/Decisions/DefensiveSetup.vue:
* Removed alignment section from template
* Removed alignment from localSetup initialization
* Removed alignmentOptions array
* Removed alignmentDisplay computed property
* Removed alignment from hasChanges comparison
* Removed alignment from visual preview (reorganized to col-span-2)
Rationale: Defensive alignment is not active in the game and will not be
used. Per Cal's decision, remove completely rather than keep as dead code.
Tests: All 728 backend unit tests passing (100%)
Session 1 Part 3 - Change #6 complete
Part of cleanup work from demo review
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Removed outdated TODO comments and updated documentation to reflect
work completed in Phase 3E (Position Ratings and WebSocket Handlers).
Changes:
- Removed 2 stale WebSocket emission TODOs in game_engine.py (lines 319, 387)
These referenced Phase 3E-Final work completed on 2025-01-10
- Updated backend/CLAUDE.md Phase 3E status section
Marked defender retrieval as COMPLETE (Phase 3E-Main)
Clarified SPD test still pending (needs batter speed rating)
Marked runner advancement as COMPLETE (Phase 3D)
- Updated TODO_RESOLUTION_SUMMARY.md
Marked defender lookup TODO as resolved with implementation details
Documentation:
- Created TODO_AUDIT_2025-01-14.md - Complete TODO audit (53 items)
- Created TODO_VERIFICATION_RESULTS.md - Verification of resolved items
- Created TODO_SUMMARY.md - Quick reference priority matrix
- Created TODO_CLEANUP_COMPLETE.md - Cleanup work summary
Test Status:
- Backend: 9/9 PlayResolver tests passing
- No regressions introduced
Remaining Work:
- 41 legitimate TODOs properly categorized for future phases
- 8 Phase F6 TODOs (game page integration)
- 5 Quick win TODOs (optional polish)
- 28 Future phase TODOs (auth, AI, advanced features)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Fixed failing test caught by pre-commit hook. The test was not properly
mocking dependencies in the resolve_play command.
Changes:
- Added mock for state_manager.get_state() to return valid state
- Added mock for random.choice() to return deterministic PlayOutcome
- Updated assertion to expect auto-generated outcome (SINGLE_1)
- Test now properly validates the auto-outcome behavior for terminal testing
Root cause: resolve_play() checks state_manager early and auto-generates
a random outcome for testing when no forced outcome is provided. Test was
not accounting for either behavior.
All 731 unit tests now passing (100%).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Test Fixes (609/609 passing):
- Fixed DiceSystem API to accept team_id/player_id parameters for audit trails
- Fixed dice roll history timing issue in test
- Fixed terminal client mock to match resolve_play signature (X-Check params)
- Fixed result chart test mocks with missing pitching fields
- Fixed flaky test by using groundball_a (exists in both batting/pitching)
Documentation Updates:
- Added Testing Policy section to backend/CLAUDE.md
- Added Testing Policy section to tests/CLAUDE.md
- Documented 100% unit test requirement before commits
- Added git hook setup instructions
Git Hook System:
- Created .git-hooks/pre-commit script (enforces 100% test pass)
- Created .git-hooks/install-hooks.sh (easy installation)
- Created .git-hooks/README.md (hook documentation)
- Hook automatically runs all unit tests before each commit
- Blocks commits if any test fails
All 609 unit tests now passing (100%)
Integration tests have known asyncpg connection issues (documented)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
