Video Processor — Claude Code skill for video, audio and Whisper
A skill that handles video files from the agent: pull the audio track, convert to mp4 or webm, transcribe speech with Whisper. Useful when video files show up in a project.
npx skills add disler/claude-code-hooks-multi-agent-observability --skill video-processor -a claude-codeWhat Video Processor does
Video Processor wraps FFmpeg and Whisper into one script with straightforward commands. Four operations are available: extract audio, convert to mp4, convert to webm, and transcribe audio or video to text. Transcription extracts the audio automatically, runs it through Whisper, and cleans up temp files.
FFmpeg and openai-whisper must be installed in the environment first. On macOS that is brew install ffmpeg, on Ubuntu apt-get install ffmpeg, and Whisper installs via pip install -U openai-whisper. The Python dependencies click and ffmpeg-python are pulled in by the script through PEP 723, and commands run with uv run.
Transcription models range from tiny to large. Bigger models give better accuracy and consume more RAM. Output formats include txt, srt, vtt, and json with word-level timestamps. For subtitles on a website, the pairing is simple: to-webm first, then transcribe to srt.
What it helps with
- pull an audio track from a video into wav, mp3, aac or flac
- convert an old camera avi to mp4
- produce a webm version for a web player
- transcribe a lecture or interview to text
- generate srt or vtt subtitles with timestamps
Example prompts for the agent
- convert old_video.avi to mp4
- extract audio from lecture.mp4 and save it as wav
- transcribe this video to srt using the small model
How to install Video Processor
1Skills CLI (Vercel)
npx skills add disler/claude-code-hooks-multi-agent-observability --skill video-processor -a claude-code2Manually: put the files in place
~/.claude/skills/video-processor/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/video-processor/
1Skills CLI (Vercel)
npx skills add disler/claude-code-hooks-multi-agent-observability --skill video-processor -a codex2Manually: put the files in place
.agents/skills/video-processor/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add disler/claude-code-hooks-multi-agent-observability --skill video-processor -a cursor2Manually: put the files in place
.cursor/skills/video-processor/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install disler/claude-code-hooks-multi-agent-observability video-processor2Manually: put the files in place
.github/skills/video-processor/SKILL.md1Skills CLI (Vercel)
npx skills add disler/claude-code-hooks-multi-agent-observability --skill video-processor -a gemini-cli2Platform CLI
gemini skills install https://github.com/disler/claude-code-hooks-multi-agent-observability --path .claude/skills/video-processorManually: put the files in place
.devin/skills/video-processor/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add disler/claude-code-hooks-multi-agent-observability --skill video-processor -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about Video Processor
How do I install Video Processor in Claude Code?
Run npx skills add disler/claude-code-hooks-multi-agent-observability --skill video-processor -a claude-code. The skill lands in ~/.claude/skills/video-processor/SKILL.md.
What does Video Processor need to work?
FFmpeg and openai-whisper must be installed in the environment. FFmpeg installs through brew or apt-get, Whisper through pip. The script will fail with a clear error if either is missing.
Which Whisper model should I pick?
Set it with the --model flag. tiny and base are fast but less accurate, small and medium balance speed and accuracy for real transcripts, large is best quality with the highest memory use. The default is base.
