video-speech-workflows — Claude Code skill
video-speech-workflows — a skill with built-in routes for trimming speech in video, adding subtitles to talking-head footage, and end-to-end editing with captions. Use when a video agent routes a single voiced video to you.
no license ★ 297 updated 2026-10-29
npx skills add wwwzhouhui/skills_collection --skill video-speech-workflows -a claude-codeWhat video-speech-workflows does
video-speech-workflows provides three modes for handling video with speech. The speech-condense mode compresses long unscripted recordings, cutting or trimming them to a target length. The talking-head-subtitles mode works with already-rendered video: adds or repairs subtitles, positions them on screen, creates bilingual lines, or marks speakers. The video-pipeline mode takes a single video and builds a full processing plan that can combine speech trimming, editing, subtitles, speaker markup, translation, or terminology notes.
This skill operates on one video file, so if you have multiple sources, assemble them first in video-edit-assembly or through the main video-edit-agent. Subtitle timings are calculated for the final video, so cues stay in sync. To inspect only a portion of video without full analysis, use video_watch_segment; reserve complete video_ingest for workflows that truly need dense contact with the entire recording.
What it helps with
- trim a long recording to a target length while preserving meaning
- add or fix subtitles on finished video
- mark speakers or create bilingual caption lines
- plan a complete edit combining speech, subtitles, and translation
Example prompts for the agent
- Compress the recording by removing pauses and cut it down to five minutes
- Add subtitles to the video with each speaker identified
- Create a full editing workflow: trim speech, sync captions, translate to French
How to install video-speech-workflows
1Skills CLI (Vercel)
npx skills add wwwzhouhui/skills_collection --skill video-speech-workflows -a claude-code2Manually: put the files in place
~/.claude/skills/video-speech-workflows/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/video-speech-workflows/
1Skills CLI (Vercel)
npx skills add wwwzhouhui/skills_collection --skill video-speech-workflows -a codex2Manually: put the files in place
.agents/skills/video-speech-workflows/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add wwwzhouhui/skills_collection --skill video-speech-workflows -a cursor2Manually: put the files in place
.cursor/skills/video-speech-workflows/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install wwwzhouhui/skills_collection video-agent-kit/skills/video-speech-workflows2Manually: put the files in place
.github/skills/video-speech-workflows/SKILL.md1Skills CLI (Vercel)
npx skills add wwwzhouhui/skills_collection --skill video-speech-workflows -a gemini-cli2Platform CLI
gemini skills install https://github.com/wwwzhouhui/skills_collection --path video-agent-kit/skills/video-speech-workflowsManually: put the files in place
.devin/skills/video-speech-workflows/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.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-speech-workflows
How do I install video-speech-workflows in Claude Code?
Run npx skills add wwwzhouhui/skills_collection --skill video-speech-workflows -a claude-code, then open the skill configuration file at ~/.claude/skills/video-speech-workflows/SKILL.md.
What does this video-speech skill do?
It provides three modes for video work: compressing speech in long recordings, adding and fixing subtitles on finished video, and building a complete editing plan that handles speech trimming, captions, and translation at once.
When should I use this skill instead of the main video agent?
This skill works with a single finished video file. If you have multiple sources or need to assemble video from clips, use video-edit-assembly or video-edit-agent first, then pass the result to this skill.