- Added warmup counter to skip first 3 steps after entering app
- Reset counter when leaving target app
- Prevents recording splash screens as videos
Co-Authored-By: Claude <noreply@anthropic.com>
- Simplified prompt to force JSON-only response
- Added debug logs to track VLM response and parsing
- Better error messages for troubleshooting
Co-Authored-By: Claude <noreply@anthropic.com>
Added missing parameters: shares, tags, category, elements
Now analysis results are properly saved to VideoRecord
Co-Authored-By: Claude <noreply@anthropic.com>
Features:
- ScreenshotAnalyzer class for VLM-based image analysis
- Real-time analysis during video recording
- Extract likes, comments, tags, category from screenshots
- Frontend display for category badges and tags
- Batch analysis API endpoint
Co-Authored-By: Claude <noreply@anthropic.com>
Features:
- VideoLearningAgent for automated video watching on Douyin/Kuaishou/TikTok
- Web dashboard UI for video learning sessions
- Real-time progress tracking with screenshot capture
- App detection using get_current_app() for accurate recording
- Session management with pause/resume/stop controls
Technical improvements:
- Simplified video detection logic using direct app detection
- Full base64 hash for sensitive screenshot change detection
- Immediate stop when target video count is reached
- Fixed circular import issues with ModelConfig
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
