agentleFS
SkillMedevSkillMedev/creator-studio/skills/thumbnail-concept/SKILL.md

thumbnail-concept

Skill6 starsChanged 38 days agoMITView on GitHub

Produces three distinct high-CTR YouTube thumbnail concepts per video - each a full spec with focal point, precise facial-expression direction, text overlay copy, color and background direction, and two paired title variants. Use when someone says "I need a thumbnail idea", "why is my CTR low", "help me pick a thumbnail and title", or is planning a video before filming and wants the click package designed first. Do NOT use for the spoken and visual opening seconds of the video itself - use video-hook-writer instead; do NOT use to write the full video - use youtube-script-writer instead; do NOT use for cropping, captioning, or exporting finished video files - use social-video-formatter instead.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

12 rules. About 0 are common, 12 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈12

    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
  • could be automatic: a setting could enforce it
  1. Distinctive:

    Every high-performing thumbnail has three elements working together, concepted together, never in isolation:

    line 23
  2. Distinctive:

    If any leg is weak, CTR drops.

    line 29
  3. Distinctive:

    Never restate the title - thumbnail and title form a two-part message, each carrying different information.

    line 39
  4. Distinctive:

    Never place text in the bottom-right third, where the duration badge covers it.

    line 41
  5. Distinctive:

    Do not put more than three visual elements in one thumbnail.

    line 67
  6. Distinctive:

    Do not use low-contrast color pairs or images that die at 120 pixels wide.

    line 68
  7. Distinctive:

    Do not let the overlay repeat the title - redundancy wastes half the click package.

    line 69
  8. Distinctive:

    Do not direct expressions that overpromise the video's actual emotional payoff.

    line 70
  9. Distinctive:

    Do not deliver one concept, and do not skip the background direction - "creator on a nice background" is not a spec.

    line 71
  10. Distinctive:

    All three concepts use different visual approaches, and each satisfies all three legs of the CTR triangle.

    line 75
  11. Distinctive:

    Every overlay is 5 words or fewer, legible at 120px, out of the duration-badge zone, and non-redundant with its title.

    line 76
  12. Distinctive:

    Every face concept has a shoot-able expression direction, not an adjective.

    line 77

≈ “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 Concept

A thumbnail is not decoration; it is the first ad for the video, and on YouTube the click package (thumbnail plus title) decides whether the video gets a chance at all - a strong video with a weak thumbnail loses to a mediocre video with a strong one. This skill produces concrete thumbnail concepts before filming or design, so the creator enters production knowing exactly what click signal they are building toward. The costly mistake it prevents is designing the thumbnail last, from whatever footage happens to exist.

Operating procedure

Step 1: gather inputs

  1. The video's core promise and emotional payoff (curiosity satisfied? problem fixed? disbelief resolved?).
  2. Whether the creator appears on camera and is willing to shoot dedicated thumbnail expressions. Default: yes if they are a face-forward channel.
  3. The channel's existing thumbnail style, so concepts either fit the brand or deliberately break pattern - name which.
  4. Any concrete visual assets: a result, a before/after, a prop, a screenshot. If none are stated, propose them and label them as suggestions.

Step 2: build each concept on the CTR triangle

Every high-performing thumbnail has three elements working together, concepted together, never in isolation:

  1. A clear focal point - a face with a strong expression, a dramatic object, or a before/after split.
  2. A text overlay of 2-5 words that adds information the image alone does not carry. Three words is the working target; five is the hard ceiling.
  3. Color contrast that pops against both YouTube's white and dark UI backgrounds.

If any leg is weak, CTR drops. A great face with redundant text, or sharp copy on a muddy image, both fail.

Step 3: direct the facial expression precisely

When a face is in frame, the expression does most of the emotional signaling. Never write "surprised" - write "mouth open, eyes wide, leaning toward camera." The expression must match the video's emotional payoff; a shocked face on a calm tutorial is a bait signal. Curiosity, disbelief, and genuine excitement are the three highest-performing emotion categories. Ban neutral expressions and posed smiles - they read as stock photography and get skipped.

Step 4: write the overlay under the hard rules

  • The overlay answers the question the image raises, or raises the question the image cannot. Image and text must not say the same thing.
  • Maximum five words; target three. Sentence case, with all-caps reserved for a single emphasis word.
  • Never restate the title - thumbnail and title form a two-part message, each carrying different information.
  • Bold, high-contrast type, legible at 120 pixels wide (mobile feed size). If it blurs at 120px, shorten it.
  • Never place text in the bottom-right third, where the duration badge covers it.

Step 5: pair two title variants per concept

For each concept, write one curiosity-gap title ("The reason your X keeps failing") and one direct-benefit title ("How to fix X in under 10 minutes"). Match deliberately: the curiosity-gap title pairs best with a high-emotion face image; the direct-benefit title pairs best with a clear result or transformation image. Deliver both so the creator can test.

Step 6: deliver three distinct concepts

Three concepts per video, each taking a different visual approach: face-forward, object/result, and text-heavy minimal. The winning thumbnail is rarely the obvious one, and a creator should never go to production with a single concept. Every concept brief includes a background color or scene direction and names where the overlay sits (upper left, center, upper right).

Worked concept spec

Video: "I raised my freelance rates 40% and lost zero clients."

Concept 1 - face-forward. Focal point: creator's face, right third of frame; expression direction: eyebrows raised, half-smile breaking, eyes at camera - "I got away with it" energy, matching the video's disbelief-to-relief payoff. Background: flat saturated teal, no scene. Overlay: "+40%. Zero lost." - upper left, white bold type with black stroke, "Zero" as the emphasis word. Contrast check: teal + white/black survives both YouTube themes. Titles: curiosity-gap - "Everyone said raising my rates would backfire"; direct-benefit - "How I raised my rates 40% without losing a client." Pairing: this high-emotion image runs with the curiosity-gap title.

Concept 2 - object/result. Focal point: two invoice screenshots side by side, old rate crossed out in red, new rate circled in green; no face. Background: dark desk blur. Overlay: "same client" - center, yellow type. Runs with the direct-benefit title.

Concept 3 - text-heavy minimal. Focal point: oversized "40%" filling the left two-thirds, red on off-white. Overlay: "and nobody left" small beneath it. Runs with the curiosity-gap title.

Deliverable

Produce three concept briefs, each containing: focal point, precise expression direction (if a face), background/scene direction, exact overlay copy with placement, a contrast note, two paired titles, and which title the concept runs with - ready to hand to a designer or to shoot against.

Do NOT

  • Do not put more than three visual elements in one thumbnail; the feed gives each image a fraction of a second, and clutter reads as noise.
  • Do not use low-contrast color pairs or images that die at 120 pixels wide.
  • Do not let the overlay repeat the title - redundancy wastes half the click package.
  • Do not direct expressions that overpromise the video's actual emotional payoff; mismatch inflates CTR once and craters retention and trust.
  • Do not deliver one concept, and do not skip the background direction - "creator on a nice background" is not a spec.

Quality bar

  • All three concepts use different visual approaches, and each satisfies all three legs of the CTR triangle.
  • Every overlay is 5 words or fewer, legible at 120px, out of the duration-badge zone, and non-redundant with its title.
  • Every face concept has a shoot-able expression direction, not an adjective.
  • Thumbnail plus title read as a two-part message in every pairing.

For the opening seconds once the click lands, pair with video-hook-writer (short-form) or youtube-script-writer (long-form) - the thumbnail's promise must be paid off inside the first 30 seconds.

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.