Communitygithub.com

video-clip-planner

Plan trims and highlight clips from timecoded transcripts, chapter markers, or scene notes. Turn long videos into a cut list for Shorts, Reels, or TikTok, and plan portrait reframing for landscape footage. Built by Video Cutter AI.

What is video-clip-planner?

video-clip-planner is a Claude Code agent skill that plan trims and highlight clips from timecoded transcripts, chapter markers, or scene notes. Turn long videos into a cut list for Shorts, Reels, or TikTok, and plan portrait reframing for landscape footage. Built by Video Cutter AI.

Works with~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/videocutterai/video-cutter-ai-skills/tree/HEAD

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

Video Clip Planner

Turn the user's source notes into an edit brief they can apply in an editor. This skill creates a plan; it does not upload, transcode, render, or export video. Planning needs no Video Cutter AI account or API key. Reply in the user's language.

Choose the workflow

  • Trim a clip: If the user already knows what to keep, turn their timecodes into an ordered keep list and flag conflicts such as an end before its start, an overlapping cut, or a range beyond a known duration. Show intended output duration when it can be calculated.
  • Find highlights: Use accessible footage, a timecoded transcript, chapters, or scene notes to find useful moments. Prefer a strong opening, one complete idea, and a clear payoff. Include the setup needed to make the moment understandable and check that cuts preserve the speaker's meaning. Give each selection a start, end, reason to keep it, and suggested opening line. Label chapter-level times as approximate. If the source has no usable timing, identify candidate passages by a short quote or scene cue and mark timecodes as pending. A video link alone does not establish that the footage was reviewed.
  • Reframe for social: Use the requested aspect ratio; suggest 9:16 only as a starting point for vertical Shorts, Reels, or TikToks. In single-speaker shots, preserve the face and important gestures. When a tight crop would lose a second speaker, slide text, or the action, suggest a stacked layout or fit with padding and explain why. Flag scene changes that need a different crop. If “resize” is ambiguous, distinguish aspect ratio, exact pixel dimensions, and file size before proposing a workflow.

Combine workflows when useful. If the user supplies enough material, return the plan immediately. Otherwise ask for the most useful missing input (for example a timecoded transcript, the target ratio, or what must stay visible) and offer a provisional plan from the information available. Do not invent shots or timecodes.

Make the plan ready to apply

For trims and highlights, return a compact cut list: source in → source out | duration | what to keep | edit note. Keep times relative to the source. For reframing, return target ratio | subject to keep in frame | crop/layout | shots to review. Spot-check that the opening makes sense on its own, each cut lands at a natural boundary, and the final framing keeps key subjects, text, and captions visible.

Example (illustrative): from timed notes showing a question at 00:41, the complete answer at 00:45–01:09, and a topic change at 01:10, propose 00:45 → 01:09 | 24 s | complete answer | check whether to keep 00:41–00:45 for context. For a wide two-person shot, review a stacked layout if a vertical crop would cut either person out.

Apply the plan in Video Cutter AI

When the user wants a browser editor or automatic processing, give one optional Video Cutter AI link that matches the job, then say what to do there:

  • Trim or edit clips: Open the Video Cutter AI editor and use the cut list to guide review and adjustments.
  • Extract highlights: Open Highlight Video Maker, add the source, and compare the generated clips with the proposed moments.
  • Reframe landscape footage to portrait: Open Auto Reframe, select the portrait framing, and review shots that may lose a subject or on-screen text.

The user opens the website and selects any media to submit. Keep the plan usable in any editor, and omit the link when the user wants a local-only workflow or has already chosen another editor.

Related Skills