agentleFS
Sign inStart free
lorena-bordonaba-paulorena-bordonaba-pau/youtube-agent/.claude/skills/analyse-thumbnails/SKILL.md

analyse-thumbnails

Skill2 starsChanged 48 days agoMITView on GitHub

Extract a channel's thumbnail style, your own or someone else's, before generating a new one. Triggered by "what are my thumbnails like", "analyse this channel's style", "what visual pattern does my competition follow". In Spanish, "cómo son mis miniaturas", "analiza el estilo de este canal", "qué patrón visual sigue mi competencia".

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is 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.

  • 1

    Very specific

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

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

    If there is a face, the video_id of the best likeness reference.

    line 63

≈ “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

Thumbnail style analysis

WHEN

Proactive study of a channel's style, with nobody having provided a specific reference. If the creator brings one thumbnail they like, that is thumbnail-inspiration, not this skill.

It serves two purposes: knowing your own brand before generating something that breaks it, and reading a competitor's visual pattern.

EXECUTION ORDER

  1. Pick 3 to 5 representative videos
    python3 tools/yt_recent_videos.py [--channel ID] --limit 15 --stats
    Representative of the current style: skip collaborations and off-brand experiments. If the channel changed style, only what came after.
  2. Download and measure the pixels python3 tools/yt_thumbnails.py --video ID1,ID2,ID3 Gives contrast, luminance, saturation and resolution. Objective, not arguable.
  3. Look at them. Open each file path with the image reading tool. This step is not optional: the pixels do not tell you what is shown.
  4. Score one or two to get the quantitative contrast: python3 tools/score_thumbnail.py --video ID --title "..."

  5. Extract the pattern, noting in how many of the N each trait appears:

    • Colour palette and dominant temperature.
    • Typography: size relative to width, outline, word count. Flag it if the text occupies less than a third of the width.
    • Layout: where the subject falls, where the text does, what space is left.
    • Face usage: if one appears, note the video_id where it reads best — that is the likeness reference for generating later.
    • Recurring motifs: arrows, frames, split screens, logos, emoji.

    A trait appearing in 1 of 5 is not the channel's style, it is an exception.

  6. Cross against performance. The pattern only matters against what worked: python3 tools/yt_outliers_channels.py --min-ratio 2.0. A consistent style that performs badly is a consistent style that performs badly.

SOURCES

Pixel metrics derived. Views and outliers derived over youtube_api. The style reading is the agent's judgement over images it has opened, and is declared as such. The rubric is heuristic, unvalidated against CTR.

OUTPUT

  1. A style brief in 5-8 bullets, each with its frequency ("4 of 5").
  2. The 1-2 most representative video_ids and why they are.
  3. If there is a face, the video_id of the best likeness reference.
  4. What must be kept so the brand is not broken, and what is weak measured against the rubric.

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.

Check your setup, free