agentleFS
ClypmintClypmint/clypmint-broll/skills/broll/SKILL.md

broll

Skill4 starsChanged 17 days agoMITView on GitHub

Source real, verified b-roll (photos and footage) for a video script with the Clypmint B-Roll API, then cut and place each shot on the word. Use when a video needs cutaways, b-roll, pictures of the people/companies/products/places a script names, or real footage of a scene.

  • Reads credentials

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 2

    1. appariciojunior/video-editor-skills★ 246same name
    2. Melvynx/video-agent-skills★ 126same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository2 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 4 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every specific person, company, product, app, website, place or event the speaker names.

    line 18
  2. Very specific:

    Always pass context: the exact line.

    line 27
  3. Distinctive:

    Always send your own plan, so the search starts on it at once and skips its own planning step:

    line 28
  4. Very specific:

    Prefer the highest score. 9-10 is usable as is.

    line 38
  5. Very specific:

    Read shows before using a shot.

    line 40
  6. Very specific:

    Keep framing as filmed. For 9:16, cut --aspect 9:16 pads (contain) by default.

    line 46
  7. Distinctive:

    Start the cutaway on (or just after) the first word of the moment, never before it.

    line 50
  8. Very specific:

    When license.attribution is set, put it in the description or end card.

    line 56
  9. Distinctive:

    Never present a picture of someone else as the named person.

    line 58
  10. Very specific:

    If usage.charged_usd is null with billing_status: "pending", billing could not be confirmed.

    line 64

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

Sourcing b-roll with Clypmint B-Roll

You have the find_broll and plan_broll tools (MCP server https://clypmint.com/mcp/broll), or the CLI npx -y github:Clypmint/clypmint-broll#v1.2.1 with CLYPMINT_API_KEY set.

1. Read the whole script first

State its point in one sentence. B-roll illustrates the point of a line, not its nearest keyword.

2. Pick the moments

In this order:

  1. Every specific person, company, product, app, website, place or event the speaker names. These have a correct shot.
  2. Concrete actions or scenes a camera could film ("we drove to the coast"). Skip opinions, feelings and abstract lines. No b-roll beats wrong b-roll. Never cover the speaker's own face at the hook.

For a whole script, call plan_broll once. It returns each moment with the exact words it starts on (and its time, if you pass word timings).

3. Search each moment well

  • query is the subject as people name it: "Sam Altman", "Notion", "Shibuya Crossing".
  • Always pass context: the exact line. It decides which "Claude" or "Apple" is meant.
  • Always send your own plan, so the search starts on it at once and skips its own planning step:
    • keywords: 1-5 archive searches of 1-3 plain words, nouns first, the most specific first. Archives match every word, so "jazz band", not "vintage jazz band performing live". Add the year or event when the line names one: ["Mandela 1990", "Mandela crowd", "Nelson Mandela"].
    • names: every named person, company, product, place or event, as Wikipedia titles it: ["Nelson Mandela"].
    • look_for: what the right shot shows, with its era, place and action: "Nelson Mandela waving to a crowd after his release from prison in 1990".
  • type: image for a named person or a logo; video for an action, a place or an event; any when unsure.
  • The real thing (a movie scene, a recent news photo): rights: "any" also searches the open web. Those shots have license.cleared: false; tell the user they are not licensed. Default is licensed.
  • Vertical edit: orientation: "portrait" ranks vertical shots first.

4. Choose

  • Prefer the highest score. 9-10 is usable as is. 7-8 is fine. Below 7, look again or drop the beat.
  • official: true means it comes from the subject's own channel or site: best for products and companies.
  • Read shows before using a shot. If it does not match the line, drop it.

5. Make the file

  • delivery: "direct": download url.
  • delivery: "local_cut" or "local_capture": run cut_command (edit --start and --duration to the moment you need). Needs ffmpeg; yt-dlp for YouTube/X/TikTok; Chrome/Edge/Brave/Chromium for captures.
  • Keep framing as filmed. For 9:16, cut --aspect 9:16 pads (contain) by default. Use --fit cover only if the user asks to crop.

6. Place it on the word

  • Start the cutaway on (or just after) the first word of the moment, never before it.
  • Hold 1.5-3 s. Cut faster in the hook, slower in the body.
  • Silent b-roll by default (--keep-audio only when its sound matters).

7. Credit and rights

  • When license.attribution is set, put it in the description or end card.
  • license.commercial_use: null (official channels, live sites): the owner holds the rights. Fine for commentary and review in most places, your call for ads.
  • Never present a picture of someone else as the named person.

Pricing to tell the user

A verified new account gets $2 of signup credit once (100 licensed or 33 web searches at regular prices), subject to the daily signup cap; then pay per search from a prepaid balance (top up from $10): $0.02 per licensed search, $0.06 per web search (rights: "any"), half price only before 2026-11-03T07:59:59Z (Nov 2, 11:59:59 p.m. Pacific), and only when the search finds verified shots. A search that finds nothing, or only near matches, costs nothing. Every answer carries usage (what it cost, the balance left, the prices now). On a 402 balance_too_low, tell the user to top up or turn on auto top-up at https://clypmint.com/broll/dashboard. broll_usage shows the balance and the prices.

If usage.charged_usd is null with billing_status: "pending", billing could not be confirmed. Check usage before retrying; do not interpret it as a zero charge.

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.