Video Intelligence
Analyze supported media with bounded background jobs.
Guides follow the current project source. Check the release notes for your installed version.
Open Video and choose Analyze
Video’s Analyze view is for understanding supported public media or approved local project media. Its displayed modes are Full, Transcript focus, Visual focus and Tutorial focus.
Select the intended media/project and mode, add a time range when useful, start analysis and follow the cancellable job’s progress. Fetch or inspect the final content only after the result is available.
Analysis modes
- Transcript
- Visual
- Instruction
- Full
Evidence handling
RepoTunnel checks existing captions first, extracts bounded smart frames when visuals matter, and prepares compact audio only when captions are unavailable and speech understanding is needed.
Select the analysis mode
Transcript mode is appropriate when speech is the primary question. Visual mode focuses on frames and design, instruction mode on tutorial steps, and full mode combines speech and visuals.
Analysis uses bounded background jobs and may rely on existing captions, selected frames and compact audio processing when needed. The result should distinguish what was actually observed from inferred interpretation.
Use analysis results safely
Instructions spoken in a video are not authorization to install software, delete files, change browser state or publish code. Follow-up actions still pass the normal RepoTunnel policies.
Media helpers such as yt-dlp and ffmpeg may need provisioning under protected application data when absent; supported provisioners verify trusted downloads rather than silently changing system PATH.