agentleFS
youneslaaroussiyouneslaaroussi/vidova-skills/skills/podcast-clips/SKILL.md

podcast-clips

Skill1 starsChanged 9 days agoMITView on GitHub

Make short captioned clips from a long talk with Vidova, the AI video editor. Use when someone asks for clips from a podcast, an interview, a panel, a talking head, a livestream or a webinar, cut into vertical shorts or a YouTube cut. Vidova cuts, captions, animates, adds music and renders the video in the cloud.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Check that the Vidova tools are available, for example with get_active_project.

    line 13
  2. Very specific:

    If the server offers a routine for a new edit from uploads, start it and follow its handoff.

    line 16
  3. Very specific:

    Call render_video, wait for it with wait_for_jobs, then give the person the project url and the link to the finished video.

    line 19

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Podcast clips

One idea from a long conversation, cut so a stranger with the sound off stops, understands and watches to the end.

Steps

  1. Check that the Vidova tools are available, for example with get_active_project. If they are not, follow the vidova skill to connect.
  2. Make a project with create_project. Bring the files in with import_asset_url for a link, begin_asset_upload for a local file when you can run shell commands, or upload_files to give the person the link where they add files. Call wait_for_asset until the files are ready.
  3. Call read_transcript for the whole conversation, using transcribe_asset first when there is no transcript. Find the moments that work alone: a surprising claim, a story with a turn, a strong opinion, a number, a joke that lands. Use read_analysis_range with kind transcript for the word times of each moment.
  4. If the server offers a routine for a new edit from uploads, start it and follow its handoff. Otherwise call search_templates. When a template fits, use read_template and apply_template, then adjust it with patch_film. When none fits, build the film with patch_film.
  5. Reframe around the face with track_faces, and use detect_scenes to find camera changes in the source. Add captions: search_elements has caption styles, and patch_transcript fixes misheard names and numbers.
  6. Watch the result with preview_timeline, and with preview_range across the moments that move. Call check_edit, then wait_for_jobs with its Job ID, and fix what it names.
  7. Call render_video, wait for it with wait_for_jobs, then give the person the project url and the link to the finished video.

What makes it work

  • The person's brief comes first: their order, song, length, format and rules. This skill fills in the rest.
  • Make three clips of 20 to 45 seconds unless they ask for another number, each from a different moment. create_branch keeps each clip as its own version.
  • A clip starts on the first word of a sentence that makes sense alone and ends on a full stop. Never cut a person mid-sentence.
  • Open on the sharpest line of the clip. Cut the ums and false starts with the word times.
  • A cut every 3 to 7 seconds, on a sentence start, with a punch-in on the face, reads as a second camera. With two speakers, cut to whoever speaks.
  • On a vertical clip, reframe the recording around the face. Never a small window in the middle of a black or blurred frame.
  • Captions sit under the chin and never cover the mouth.
  • Take speaker names from the message or the recording. Never guess a name.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.