- Add sidecar management to backend (sidecar_manager.py)
- Add sidecar API router for browser mode (/api/sidecar/*)
- Add browser-api.js polyfill for running in Chrome/Edge
- Add "Open in Browser" button when audio access fails
- Update build scripts with new sidecar modules
- Add start-browser.sh for development browser mode
Browser mode allows users to open the app in their system browser
when Electron's audio access is blocked by security software.
The backend manages the sidecar process in browser mode (BROWSER_MODE=true).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add a new collapsible panel in meeting-detail page for audio device
management with the following features:
- Device dropdown selector showing all available microphones
- Real-time volume meter using Web Audio API AnalyserNode
- Test recording function (5 seconds max with countdown)
- Test playback function to verify recording quality
- Device hot-plug detection
- Preference persistence in localStorage
- Traditional Chinese localization
The main recording function now uses the user-selected device instead
of auto-selecting, giving users full control over which microphone
to use for meeting transcription.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add isAlias() helper to identify 'default' and 'communications' aliases
- Prefer actual device IDs (long strings) over alias IDs
- For alias deviceIds, use {audio: true} to let system choose
- For real deviceIds, try exact first, then ideal as fallback
- Add debug logging for isSecureContext and alias detection
This fixes the "Could not start audio source" error when Electron tries
to open a microphone using exact: 'communications' constraint.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add more console.log statements to help debug sidecar startup issues:
- Log when startSidecar() is called
- Log sidecar directory and packaged status
- Log executable path check results
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Set sidecarReady to false when model_error is received
- Store error message in activeWhisperConfig for status display
- Update frontend to show error state with red indicator
- Prevents showing ✅ when model failed to load
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Remove icon references from package.json (assets folder is empty)
- Fix batch script syntax by replacing if/else blocks with goto labels
- Remove parentheses and colons that conflict with batch syntax
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add NSIS installer as default target (--target nsis)
- Keep portable option available (--target portable)
- Store SQLite database in %APPDATA%\Meeting-Assistant for persistence
- Portable temp folder cleanup no longer affects SQLite data
- Update build-client.bat with --target parameter support
- Update DEPLOYMENT.md with new options and comparisons
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add robust model cache verification (check model.bin + config.json)
- Add new status messages: model_cached, incomplete_cache, model_error
- Forward model status events to frontend for better UI feedback
- Add clean_build_cache() to remove stale spec files before build
- Add --clean flag to PyInstaller commands
- Change sidecar from --onefile to --onedir for faster startup
- Add missing hidden imports: onnxruntime, wave, huggingface_hub.utils
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Disable AudioServiceOutOfProcess feature which can cause audio capture issues
- Enable WebRTCPipeWireCapturer for better audio support
- Set autoplay-policy to no-user-gesture-required
These flags must be set before app is ready and help resolve
"Could not start audio source" errors in Electron while the
same microphone works fine in regular browsers.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Filter out "立體聲混音" (Stereo Mix) devices which are not real microphones
- Skip the "default" device which may point to Stereo Mix
- Prefer "communications" device or devices with "麥克風/microphone" in label
- Use exact deviceId constraint for better device selection
- Add fallback to preferred constraint if exact fails
- Add more detailed console logging for debugging
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add device enumeration check before attempting to capture audio
- Use simpler audio constraints (audio: true) instead of specific options
- Add fallback to explicit device ID if simple constraints fail
- Add more descriptive error messages for different failure modes
- Enhance Electron permission handlers with better logging
- Add setDevicePermissionHandler for audio device access
- Include 'microphone' in allowed permissions list
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Added session.setPermissionRequestHandler and setPermissionCheckHandler
to automatically grant media/audio capture permissions. This fixes the
"could not start audio source" error when trying to start recording.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add SQLite as alternative database for offline/firewall environments
- Add --database-type parameter to build-client.bat (mysql/sqlite)
- Refactor database.py to support both MySQL and SQLite
- Add DB_TYPE and SQLITE_PATH configuration options
- Set fixed unpackDirName for portable exe (Meeting-Assistant)
- Update DEPLOYMENT.md with SQLite mode documentation
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- main.js: Strip UTF-8 BOM when reading config.json
- build-client.bat: Write config.json without BOM using
[System.Text.UTF8Encoding]::new($false)
PowerShell's -Encoding UTF8 writes BOM by default, which can
cause JSON parsing issues in Node.js.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The embedded flag must be true for the portable exe to work.
When embedded is false, the app expects an external backend.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add backend/run_server.py entry point for embedded deployment
- Add backend/build.py PyInstaller script for backend packaging
- Modify config.py to support frozen executable paths
- Extend client/config.json with backend configuration section
- Add backend sidecar management in Electron main process
- Add Whisper model download progress reporting
- Update build-client.bat with --embedded-backend flag
- Update DEPLOYMENT.md with all-in-one deployment documentation
This enables packaging frontend and backend into a single executable
for simplified enterprise deployment. Backward compatible with
existing separate deployment mode (backend.embedded: false).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add activeWhisperConfig tracking in main.js to expose current Whisper settings
- Expand get-sidecar-status IPC handler to return whisper config (model, device, compute, configSource)
- Add Whisper status display in meeting-detail.html transcript panel header
- Status updates every 5 seconds and shows: model, device, compute type, and ready state
- Add comprehensive debug logging for config loading and whisper config resolution
- Helps diagnose why config.json settings may not be passed correctly to sidecar
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add signAndEditExecutable: false to skip code signing on Windows
- Clear electron-builder winCodeSign cache before build
- Add helpful error messages for symlink permission issues
The winCodeSign package contains macOS symlinks that fail to extract
on Windows without admin privileges or Developer Mode enabled.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Backend:
- Add setup-backend.sh/bat for one-click backend setup
- Fix test_auth.py mock settings (JWT_EXPIRE_HOURS)
- Fix test_excel_export.py TEMPLATE_DIR reference
Frontend:
- Add config.json for runtime API URL configuration
- Add init.js and settings.js for config loading
- Update main.js to load config from external file
- Update api.js to use dynamic API_BASE_URL
- Update all pages to initialize config before API calls
- Update package.json with extraResources for config
Build:
- Add build-client.sh/bat for packaging Electron + Sidecar
- Add build-all.ps1 PowerShell script with -ApiUrl parameter
- Add GitHub Actions workflow for Windows builds
- Add scripts/README.md documentation
This allows IT to configure backend URL without rebuilding.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add audio file upload transcription via Dify STT API
- Implement VAD-based audio segmentation in sidecar (3-min chunks)
- Add SSE endpoint for real-time transcription progress updates
- Fix chunk size enforcement for reliable uploads
- Add retry logic with exponential backoff for API calls
- Support Python 3.13+ with audioop-lts package
- Update frontend with Chinese progress messages and chunk display
- Improve start.sh health check with retry loop
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Sync DOM values to currentMeeting object before re-rendering when user
clicks Add or Remove buttons. Previously, typed content was lost because
render functions read from the data model which wasn't updated.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Enterprise Meeting Knowledge Management System with:
Backend (FastAPI):
- Authentication proxy with JWT (pj-auth-api integration)
- MySQL database with 4 tables (users, meetings, conclusions, actions)
- Meeting CRUD with system code generation (C-YYYYMMDD-XX, A-YYYYMMDD-XX)
- Dify LLM integration for AI summarization
- Excel export with openpyxl
- 20 unit tests (all passing)
Client (Electron):
- Login page with company auth
- Meeting list with create/delete
- Meeting detail with real-time transcription
- Editable transcript textarea (single block, easy editing)
- AI summarization with conclusions/action items
- 5-second segment recording (efficient for long meetings)
Sidecar (Python):
- faster-whisper medium model with int8 quantization
- ONNX Runtime VAD (lightweight, ~20MB vs PyTorch ~2GB)
- Chinese punctuation processing
- OpenCC for Traditional Chinese conversion
- Anti-hallucination parameters
- Auto-cleanup of temp audio files
OpenSpec:
- add-meeting-assistant-mvp (47 tasks, archived)
- add-realtime-transcription (29 tasks, archived)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
