Add a new "Business" theme designed for corporate/professional use cases.
The theme features a sophisticated navy and gray color palette that conveys
trust and professionalism while maintaining excellent readability.
Key characteristics:
- Deep navy primary color for trust and professionalism
- Off-white/charcoal backgrounds (avoiding harsh pure white/black)
- Teal accent for CTAs and highlights
- Soft, professional shadows
- System fonts for native feel
- High contrast ratios (WCAG AA compliant)
Files changed:
- globals.css: Added .theme-business light and dark mode variables
- useTheme.ts: Added 'business' to ThemeId and THEMES array
- ThemeSelector.tsx: Added business theme class handling
The comment said "excluding test files" but the grep commands didn't
actually exclude them. Added common test file exclusion patterns.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add isComposing check to prevent Enter key from submitting messages
while Japanese (or other) IME input is in progress.
Affected components:
- AssistantChat
- ExpandProjectChat
- SpecCreationChat
- FolderBrowser
- TerminalTabs
Added 'text' language identifier to all fenced code blocks in the
Infrastructure Feature Descriptions section to satisfy MD040.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Addresses reviewer feedback:
1. Windows Compatibility: Added Windows alternative using netstat/taskkill
2. Safer Process Killing: Changed from `pkill -f "node"` to port-based
killing (`lsof -ti :$PORT`) to avoid killing unrelated Node processes
like VS Code, Claude Code, or other development tools
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Changed grep for "new Map()/new Set()" to search all of src/
- Previously only searched src/lib/, src/store/, src/data/
- Now consistent with other grep patterns that search entire src/
- Applied to both coding_prompt and initializer_prompt templates
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Changes:
- Replace pkill -f "node" with port-based killing (lsof -ti :PORT)
- Safer: only kills dev server, not VS Code/Claude Code/other Node apps
- More specific: targets exact port instead of all Node processes
- Add Windows alternative commands (commented, for reference)
- Use ${PORT:-3000} variable instead of hardcoded port 3000
- Update health check and API verification to use PORT variable
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add a comprehensive theme system allowing users to switch between three
distinct visual themes, each supporting both light and dark modes:
- Twitter (default): Clean blue design with soft shadows
- Claude: Warm beige/cream tones with orange primary accents
- Neo Brutalism: Bold colors, hard shadows, 0px border radius
New files:
- ui/src/hooks/useTheme.ts: Theme state management hook with localStorage
persistence for both theme selection and dark mode preference
- ui/src/components/ThemeSelector.tsx: Header dropdown with hover preview
and color swatches for quick theme switching
Modified files:
- ui/src/styles/globals.css: Added CSS custom properties for Claude and
Neo Brutalism themes with light/dark variants, shadow variables
integrated into @theme inline block
- ui/src/App.tsx: Integrated useTheme hook and ThemeSelector component
- ui/src/components/SettingsModal.tsx: Added theme selection UI with
preview swatches and dark mode toggle
- ui/index.html: Added DM Sans and Space Mono fonts for Neo Brutalism
Features:
- Independent theme and dark mode controls
- Smooth CSS transitions when switching themes
- Theme-specific shadow styles (soft vs hard)
- Theme-specific fonts and border radius
- Persisted preferences in localStorage
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Allow Foundry and custom deployments to override the default Opus model
by setting the ANTHROPIC_DEFAULT_OPUS_MODEL environment variable.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Remove translate-x/translate-y CSS selectors that broke layout utilities
(AssistantPanel slide animation, DebugLogViewer resize handle)
- Add browser validation to get_playwright_browser() with warning for
invalid values (matches get_playwright_headless() behavior)
- Remove phantom SQLite documentation from CUSTOM_UPDATES.md that
described features not present in PR #93
- Update checklist and revert instructions to match actual changes
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Check for 'not found' message (server response) in addition to '404'
- Only clear stored conversation ID on actual 404 errors
- Prevent unnecessary retries for deleted conversations
- Don't clear conversation on transient network errors
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
When a stored conversation ID no longer exists (e.g., database was reset
or conversation was deleted), the UI would repeatedly try to fetch it,
causing endless 404 errors in the console.
This fix:
- Stops retrying on 404 errors (conversation doesn't exist)
- Automatically clears the stored conversation ID from localStorage
when a 404 is received, allowing the user to start fresh
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Remove assets/ollama.png (duplicate of ui/public/ollama.png)
- Remove .claude/settings.local.json from tracking
- Add .claude/settings.local.json to .gitignore
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
pkill on BSD systems accepts multiple pattern operands. Previous code
only validated args[-1], allowing disallowed processes to slip through
when combined with allowed ones (e.g., "pkill node sshd" would only
check "sshd").
Now validates every non-flag argument to ensure no disallowed process
can be targeted. Added tests for multiple pattern scenarios.
Addresses CodeRabbit feedback on PR #101.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Changed from warning-only to proper error handling:
- if server doesn't respond after restart, exit with error
- prevents false negatives when server fails to start
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Address CodeRabbit security feedback - restrict pkill_processes entries
to alphanumeric names with dots, underscores, and hyphens only.
This prevents potential exploitation through regex metacharacters like
'.*' being registered as process names.
Changes:
- Added VALID_PROCESS_NAME_PATTERN regex constant
- Updated both org and project config validation to:
- Normalize (trim whitespace) process names
- Reject names with regex metacharacters
- Reject names with spaces
- Added 3 new tests for regex validation
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Address CodeRabbit feedback - use consistent conditional pattern:
`(max_priority.priority + 1) if max_priority else 1`
This matches the pattern used in create_feature and create_features_bulk.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Users can now access the WebUI remotely (e.g., via VS Code tunnels,
remote servers) by specifying a host address:
python start_ui.py --host 0.0.0.0
python start_ui.py --host 0.0.0.0 --port 8888
Changes:
- Added --host and --port CLI arguments to start_ui.py
- Security warning displayed when remote access is enabled
- AUTOCODER_ALLOW_REMOTE env var passed to server
- server/main.py conditionally disables localhost middleware
- CORS updated to allow all origins when remote access is enabled
- Browser auto-open disabled for remote hosts
Security considerations documented in warning:
- File system access to project directories
- API can start/stop agents and modify files
- Recommend firewall or VPN for protection
Fixes: leonvanzyl/autocoder#81
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
When running multiple projects simultaneously, UI would show mixed data
because the manager registry used only project_name as key. Projects with
the same name but different paths shared the same manager instance.
Changed manager registries to use composite key (project_name, resolved_path):
- server/services/process_manager.py: AgentProcessManager registry
- server/services/dev_server_manager.py: DevServerProcessManager registry
This ensures that:
- /old/my-app and /new/my-app get separate managers
- Multiple browser tabs viewing different projects stay isolated
- Project renames don't cause callback contamination
Fixes: leonvanzyl/autocoder#71
Also fixes: leonvanzyl/autocoder#62 (progress bar sync)
Also fixes: leonvanzyl/autocoder#61 (features missing in kanban)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
When all features pass, the orchestrator continued spawning testing
agents for 10+ minutes, wasting tokens on unnecessary regression
tests. Added a check for get_all_complete() to prevent this.
Fixes: leonvanzyl/autocoder#66
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The REST API skip endpoint was using max_priority + 1000, while the
MCP server used max_priority + 1. This caused priority inflation where
values could reach 10,000+ after multiple skips.
Changed to use + 1 for consistency with mcp_server/feature_mcp.py:345.
Fixes: leonvanzyl/autocoder#65
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Escape parentheses in grep patterns: new Map\(\) and new Set\(\)
- Add --include="*.js" to all grep commands for complete coverage
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Fix math error in category totals (155→165, 255→265)
- Fix example JSON to include [0,1,2,3,4] dependencies for all features
- Add more robust server shutdown (SIGTERM then SIGKILL)
- Add health check after server restart
- Align grep patterns between templates (add .js, testData, TODO/STUB/MOCK)
- Add package.json check for mock backend libraries
- Reference STEP 5.6 instead of duplicating grep commands
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add focus-visible styles for keyboard navigation accessibility and
improve PLAYWRIGHT_HEADLESS environment variable validation to warn
users about invalid values instead of silently defaulting.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Problem:
Several API endpoints return 500 Internal Server Error because datetime
objects are not serializable by Pydantic. The error occurs when:
- GET /agent/{project}/status
- GET /devserver/{project}/status
- GET /schedules/{project}/next
Root cause:
Pydantic models expect strings for Optional datetime fields, but the code
was passing raw datetime objects.
Solution:
Convert datetime objects to ISO 8601 strings using .isoformat() before
returning in Pydantic response models.
Changes:
- server/routers/agent.py: Fix started_at serialization
- server/routers/devserver.py: Fix started_at serialization
- server/routers/schedules.py: Fix next_start/next_end serialization
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Problem:
The coding agent can implement in-memory storage (e.g., `dev-store.ts` with
`globalThis`) instead of a real database. These implementations pass all tests
because data persists during runtime, but data is lost on server restart.
This is a root cause for #68 - agent "passes" features that don't actually work.
Solution:
1. Add 5 mandatory Infrastructure Features (indices 0-4) that run first:
- Feature 0: Database connection established
- Feature 1: Database schema applied correctly
- Feature 2: Data persists across server restart (CRITICAL)
- Feature 3: No mock data patterns in codebase
- Feature 4: Backend API queries real database
2. Add STEP 5.7: Server Restart Persistence Test to coding prompt:
- Create test data, stop server, restart, verify data still exists
3. Extend grep patterns for mock detection in STEP 5.6:
- globalThis., devStore, dev-store, mockData, fakeData
- TODO.*real, STUB, MOCK, new Map() as data stores
Changes:
- .claude/templates/initializer_prompt.template.md - Infrastructure features
- .claude/templates/coding_prompt.template.md - STEP 5.6/5.7 enhancements
- .claude/commands/create-spec.md - Phase 3b database question
Backwards Compatible:
- Works with YOLO mode (uses bash/grep, not browser automation)
- Stateless apps can skip database features via create-spec question
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
UI Changes:
- Replace neobrutalism with clean Twitter/Supabase-style design
- Remove all shadows, use thin borders (1px)
- Single accent color (Twitter blue) for all status indicators
- Rounded corners (1.3rem base)
- Fix dark mode contrast and visibility
- Make KanbanColumn themeable via CSS classes
Backend Changes:
- Default Playwright browser changed to Firefox (lower CPU)
- Default Playwright mode changed to headless (saves resources)
- Add PLAYWRIGHT_BROWSER env var support
Documentation:
- Add CUSTOM_UPDATES.md with all customizations documented
- Update .env.example with new Playwright options
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Create custom-theme.css for theme overrides that won't conflict
with upstream updates. The file loads after globals.css, so its
CSS variables take precedence.
This approach ensures:
- Zero merge conflicts on git pull (new file, not in upstream)
- Theme persists across upstream updates
- Easy to modify without touching upstream code
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Added _kill_process_tree call in _read_output finally block to ensure
child processes (Claude CLI) are cleaned up when agents complete or fail.
This prevents accumulation of zombie processes that was causing 78+
Python processes when max concurrency was set to 5.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Resolved conflicts by combining:
- stdin=DEVNULL and CREATE_NO_WINDOW (blocking fix)
- PYTHONUNBUFFERED env var (output buffering fix)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add stdin=subprocess.DEVNULL to prevent blocking on stdin reads
- Add CREATE_NO_WINDOW flag on Windows to prevent console pop-ups
- Remove trailing pause from start_ui.bat
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Remove the testing_in_progress claim/release mechanism from the testing
agent architecture. Multiple testing agents can now test the same feature
concurrently, simplifying the system and eliminating potential stale lock
issues.
Changes:
- parallel_orchestrator.py:
- Remove claim_feature_for_testing() and release_testing_claim() methods
- Remove _cleanup_stale_testing_locks() periodic cleanup
- Replace with simple _get_random_passing_feature() selection
- Remove startup stale lock cleanup code
- Remove STALE_TESTING_LOCK_MINUTES constant
- Remove unused imports (timedelta, text)
- api/database.py:
- Remove testing_in_progress and last_tested_at columns from Feature model
- Update to_dict() to exclude these fields
- Convert _migrate_add_testing_columns() to no-op for backwards compat
- mcp_server/feature_mcp.py:
- Remove feature_release_testing tool entirely
- Remove unused datetime import
- prompts.py:
- Update testing prompt to remove feature_release_testing instruction
- Testing agents now just verify and exit (no cleanup needed)
- server/websocket.py:
- Update AgentTracker to use composite keys (feature_id, agent_type)
- Prevents ghost agent creation from ambiguous [Feature #X] messages
- Proper separation of coding vs testing agent tracking
Benefits:
- Eliminates artificial bottleneck from claim coordination
- No stale locks to clean up after crashes
- Simpler crash recovery (no testing state to restore)
- Reduced database writes (no claim/release transactions)
- Matches intended design: random, concurrent regression testing
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
