Video understanding of DeepSeek Harness
dundunhan/dsh-video-lens · skills
Video understanding for DeepSeek Harness — give text-only agents eyes and ears on video.
★ 117GitHub stars
2Forks
2026-08-23Last updated
JavaScriptLanguage
MITLicense
Key features
- Scene changes are detected with ffmpeg's scdet filter (ffmpeg ≥ 6.0).
- Readsany local file path the agent passes to its tools (via ffprobe/ffmpeg).
- Videos without detectable cuts fall back to uniform midpoint sampling.
- ASR is strictly additiveif asrApiKeyEnv is unset or the provider fails, the visual analysis still completes and transcript is null.
- All media work is delegated to ffmpeg/ffprobe on PATH — no native decoding in the agent.
Requirements
- Tested with DSH profile bundles @deepseek-ai/dsh-base + @deepseek-ai/dsh-web-app.
- Node ≥ 20 (uses AbortSignal.any / built-in fetch / FormData).
Install command
dsh plugin --profile web add github:dundunhan/dsh-video-lens