agentleFS
aya-talebaya-taleb/video-boost-skills/skills/similar-videos/SKILL.md

similar-videos

Skill1 starsChanged 23 days agoMITView on GitHub

Finds the top videos most similar to yours (same topic, format, platform, market and language), collects their public facts (title, channel, date, length, views with source, chapters, description, hashtags), and compares them with your video side by side, so you see what the winners do that you don't. Use when the user says compare my video with similar ones, who are my competitors, what do the top videos on this do, or as part of video-score.

  • Reads credentials

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Never guess a view count. Channel stats sites give estimates.

    line 29
  2. Distinctive:

    Every fact has a link. Every view count has a source and a date.

    line 39

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Similar videos

A score against a checklist says how well you follow the rules. This skill says how you stand against the real videos you compete with.

Inputs

InputDefault
your video, and its title and description if it has themask
topic and format (for example "Atomic Habits, animated book summary")from the video
platformyoutube
market, languageUS, en
window12 months for "recent"; up to 3 older all-time leaders are allowed and marked older

Steps

  1. Find candidates.
    • YouTube with YOUTUBE_API_KEY: python3 ../trend-ideas/scripts/trend_scan.py --angles "<topic>" "<topic> summary" --region US --lang en --days 365 --long --out similar.json. It adds the outlier score (views vs the channel's median of its last 10 uploads), so you can tell a video that won on its own from one that rode a big channel. For older leaders run ../shorts-research/scripts/top_shorts.py --topic "<topic>" --long --months 120.
    • Otherwise: search the web for the topic plus the format and the year, and open the video pages.
    • TikTok and Reels: public pages only, no logins, no scraping tools.
  2. Pick 5 to 10. Same topic, or the same format on a close topic. Prefer videos from the last 12 months. Up to 3 older leaders, marked older. Drop videos in another language or made for another market (say why).
  3. Collect public facts for each, from the video page or a dated stats page: title · channel · post date · length · views (with the source and the date you read it; leave empty if no page shows it) · outlier (if scanned) · channel subscribers · chapters (count) · first 2 lines of the description · links in the description · hashtags · title pattern · thumbnail (describe it only if you can see it). Never guess a view count. Channel stats sites give estimates; say so.
  4. Do not download other people's videos. Platform terms do not allow it. If the user wants pacing or hook details, ask them to watch the first 30 seconds and tell you, or to share screenshots.
  5. Compare. One table: your video vs each similar video, one row per fact. Then:
    • What they all do that you don't (only patterns seen in 3 or more of them)
    • Where you are different in a good way (a gap you can own)
    • What you could not compare and why
  6. Save similar-videos.json (the facts) and similar-videos.md (the table and findings). video-packaging and video-score read them.

Rules

  • Every fact has a link. Every view count has a source and a date.
  • Patterns need 3 or more videos. One video is an example, not a pattern.
  • Plain words. Short lines.

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.