Claude Skill MartBrowse skillsQuick linesLearn by videoTerminal guideWhat is a Skill?
 
← Back to list

video-cut

Executes an agent's editing decisions: validates clip_plan.json, snaps cut points to safe scene/sentence boundaries, and concatenates the source video into edited_source.mp4.

Image & VideoAdvanced★ 470⑂ 86AI score 8/10Last updated: Aug 24, 2026

What it does

  • Validates and normalizes clip_plan.json (source-time start/end ranges) into clip_plan_validated.json with clip_id, source/output times and durations.
  • Concatenates the chosen ranges into edited_source.mp4. Deterministic output; the render is reused when it is newer than the clip plan.
  • Enforces cut-boundary safety: it first snaps to nearby hard cuts to avoid flash frames (SCENE_CUT_SNAP, margin 0.5s, detect threshold 0.4), then snaps to reliable sentence ends / silence windows. If ASR says the in/out point is mid-speech and cannot be snapped, it blocks with unsafe_clip_sentence_boundary.
  • Overlapping/duplicate source ranges are rejected by default; --allow-overlap requires narration lines to carry source_clip_id. Same-source continuous joins are lossless with no mid-sentence fades; non-continuous joins get anti-pop fades inside safe pauses.
  • Legacy single-stage mode also maps source-time narration into narration_mapped.json; the orchestrated flow uses --no-narration-map and stops after the cut.

Who it's for

  • Builders of agent-driven recap / video-summary pipelines who want the LLM to make creative choices while a script handles technically safe cutting.
  • Editors automating ffmpeg-based assembly who keep hitting half-sentence audio and jarring joins.
  • Intermediate-to-advanced users comfortable with Python + ffmpeg CLI; the docs are written in Chinese.

Examples

  1. With ASR data and speech_boundary_anchors.json from the understanding phase, the agent drafts rough ranges, then runs python3 scripts/cut.py movie.mp4 --work-dir work --target-duration 10m --no-narration-map and writes narration.json against the output timeline.
  2. To audit rapid cutting, recall candidates with ffmpeg -i input.mp4 -vf "select='gt(scene,0.35)',showinfo" -an -f null -, then classify them as original short shots, flash-like shots needing extended IN/OUT, or artifacts of this concatenation.
  3. To intentionally reuse a source moment twice, enable --allow-overlap and tag each narration line with source_clip_id for disambiguation.

· · · Install guide · · ·

Try it now, no install

Paste this into Claude to use the skill without installing anything.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/zenstory-ai/video-recap-skills/HEAD/skills/video-cut/SKILL.md

What I want: (describe your task here)

If Claude can't open the link, open it yourself and paste the contents instead.

↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the skills/video-cut folder from the GitHub repo zenstory-ai/video-recap-skills into my ~/.claude/skills/zenstory-ai-video-cut/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/zenstory-ai/video-recap-skills.git && mkdir -p ~/.claude/skills && cp -r video-recap-skills/skills/video-cut ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Open a terminal in your working folder.
  2. Clone the repo: git clone https://github.com/zenstory-ai/video-recap-skills.git
  3. Create the skills folder: mkdir -p ~/.claude/skills
  4. Copy this skill: cp -r video-recap-skills/skills/video-cut ~/.claude/skills/
  5. Verify prerequisites: python3 --version and ffmpeg -version must both work (macOS: brew install ffmpeg).
  6. Prepare a work_dir containing the source video and a clip_plan.json. This skill never picks clips itself, so also install the repo's understanding/planning skills for the full two-stage flow.
  7. Restart Claude Code and ask something like "cut the video using this clip plan" to trigger the skill.
View source on GitHub ↗License: MIT