thumbnail-design — Claude Code skill: YouTube thumbnail briefs
This skill writes the thumbnail brief and the A/B test plan: one focal point, mobile-legible text, an honest promise instead of clickbait.
npx skills add social-media-skills/skills --skill thumbnail-design -a claude-codeWhat thumbnail-design does
The skill treats a thumbnail as a promise, not a poster. It reads brand-profile plus the video and its title first, then writes a one-sentence promise and works the concept through the CLICK framework: one focal point, genuine emotion, contrast, short text, testing. Thumbnail and title are designed as one unit — the image shows, the title tells, and neither repeats the other.
What you get is a brief, not a raster. A human or an image tool makes the final image: nano-banana, ideogram, Canva, a designer. Uploading the custom YouTube thumbnail and running Test & Compare stay native in Studio, since WoopSocial has no thumbnail field. TikTok and IG-Reel covers go another way — they use WoopSocial's cover field through the format skills.
The rules are strict. Clickbait is out because the native test picks a winner by watch-time share, not raw CTR, so a misleading thumbnail loses to an honest one. Unconsented real-person faces and deepfakes are off the table, and the skill never fabricates a CTR — real testing and native analytics only. Specs: 1280×720, 16:9, 0–4 bold words with an outline, legible at roughly 168×94.
What it helps with
- write the thumbnail promise in one sentence
- pick thumbnail text that doesn't duplicate the title
- produce a brief for a designer or image generator
- plan an A/B test of three different concepts
- check a mockup for mobile legibility
Example prompts for the agent
- write a thumbnail brief for this video, title attached
- pick the thumbnail text and check it doesn't repeat the title
- build an A/B test plan: three concepts, one variable
How to install thumbnail-design
1Skills CLI (Vercel)
npx skills add social-media-skills/skills --skill thumbnail-design -a claude-code2Manually: put the files in place
~/.claude/skills/thumbnail-design/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/thumbnail-design/
1Skills CLI (Vercel)
npx skills add social-media-skills/skills --skill thumbnail-design -a codex2Manually: put the files in place
.agents/skills/thumbnail-design/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add social-media-skills/skills --skill thumbnail-design -a cursor2Manually: put the files in place
.cursor/skills/thumbnail-design/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install social-media-skills/skills thumbnail-design2Manually: put the files in place
.github/skills/thumbnail-design/SKILL.md1Skills CLI (Vercel)
npx skills add social-media-skills/skills --skill thumbnail-design -a gemini-cli2Platform CLI
gemini skills install https://github.com/social-media-skills/skills --path skills/thumbnail-designManually: put the files in place
.devin/skills/thumbnail-design/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add social-media-skills/skills --skill thumbnail-design -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about thumbnail-design
How do I install thumbnail-design in Claude?
Run npx skills add social-media-skills/skills --skill thumbnail-design -a claude-code. The skill lands in ~/.claude/skills/thumbnail-design/SKILL.md.
What does thumbnail-design do?
It writes the thumbnail promise, concept and text, then produces a brief and an A/B test plan using the CLICK framework. It does not render the raster.
Who creates the final image?
A human or an image tool: nano-banana, ideogram, Canva, a designer. The YouTube upload and Test & Compare run natively in Studio.
Source text
SKILL.md · MIT
--- name: thumbnail-design description: >- Design YouTube video thumbnails that earn the click — mobile-legible, one focal point, honest. Use when someone wants to design a YouTube thumbnail, improve CTR, brief a thumbnail, choose thumbnail text, or plan thumbnail A/B testing. Designs and briefs high-CTR thumbnails and writes the A/B test plan: a thumbnail is a promise, not a poster, designed as a unit with the title and legible on mobile. Uses the CLICK framework. Reads brand-profile + the video/title first. A HUMAN or an image/design tool (nano-banana/ideogram/Canva/designer) creates the final raster; the YouTube custom thumbnail is uploaded NATIVELY in Studio (WoopSocial has no thumbnail field) and Test & Compare is native; TikTok/IG-Reel covers use WoopSocial's cover field via those format skills. No clickbait, no unconsented real-person/deepfake faces, CTR never fabricated. Distinct from youtube-publishing-and-metadata (the metadata) and nano-banana/ideogram (which generate the raster this briefs). version: 1.0.0 --- # thumbnail-design A **design-craft skill** — write the promise, design the concept + text, produce the **brief**, and plan the **A/B test**. The agent **designs + briefs + plans**; a **human or image tool makes the raster**; the **YouTube thumbnail is uploaded + tested natively** (no WoopSocial field). ## The POV: a promise, not a poster — designed as a unit with the title A thumbnail drives **~70–80% of click decisions**, and the viewer decides in **under a second** — so its job is the **click that leads to a watch.** Two rules anchor everything: the **thumbnail + title are one unit** (the thumbnail **shows**, the title **tells** — they never repeat), and it's a **promise, not a poster.** That second rule is enforced by the platform: YouTube's native A/B test picks the winner by **watch-time share, not raw CTR**, so a misleading high-click thumbnail **loses** to an honest one. Design for **mobile first** — most viewing is on a phone, where simplicity and contrast win. ## Read these first 1. **brand-profile** — channel style/template (for consistency). 2. the **video and its title** (often from **youtube-publishing-and-metadata**/**youtube-long-form**/ **youtube-shorts**). ## The framework: CLICK (Depth: `references/the-click-framework.md`.) - **C — Clarify the promise:** one-sentence promise; thumbnail + title as a unit (show vs tell, never repeat); promise not poster. - **L — Lead with one focal point:** one/two hero elements, minimal background, parseable in <1s; one directional element. - **I — Invoke genuine emotion:** expressive genuine face + gaze if it adds context (skip it otherwise); match the expression to the content. - **C — Contrast + concise text:** complementary high contrast (avoid YouTube red/white, 60-30-10, B&W test, light+dark); 0–4 bold sans-serif words + outline, not a title duplicate; mobile-legible at ~168×94; safe zones; 1280×720, 16:9. - **K — Keep testing:** 3 meaningfully-different concepts, one variable; native Test & Compare picks by watch-time share (not Shorts, needs traffic); a human uploads + runs it; keep a log; brand-consistent template. ## The reality (verify-quarterly) Promise-not-poster + thumbnail+title-as-unit + watch-time A/B punishes clickbait; the three principles (simplicity/contrast/emotion); faces (+20–30%, genuine, gaze, unknown-face-can-hurt); contrast/color (complementary, avoid YT red/white, 60-30-10, B&W test); text (0–4 words, bold sans-serif, no title duplicate); mobile-first + safe zones + specs (1280×720; 50MB cap since early 2026); brand consistency; native Test & Compare (watch-time-share winner, also tests titles/combos, not Shorts, traffic-dependent): `references/thumbnail-2026-reality.md`. The promise template, design checklist, spec block, A/B plan + worked examples: `references/design-and-test-recipes.md`. ## Honest scope (never violate) - **The agent** writes the promise/concept/text/**brief** and can **critique a mockup** — but a **HUMAN or image tool** (`nano-banana`/`ideogram`/Canva/a designer) **creates the final raster.** - **The YouTube custom thumbnail is uploaded NATIVELY in Studio — WoopSocial has no thumbnail field** — and **Test & Compare is native** (a human uploads + runs it). **TikTok/IG-Reel covers** are different: those use **WoopSocial's `cover` field** via `tiktok-video-publishing`/the IG skill. - **No clickbait** (the thumbnail must keep the promise — the watch-time test punishes deception); **no unconsented real-person/deepfake faces**; **never fabricate a CTR** (real testing/native analytics only). (Scope, distinctions + connections: `references/scope-and-connections.md`.) ## Distinct from its siblings (route correctly) **thumbnail-design (this)** = the thumbnail design craft + brief + A/B plan · **youtube-publishing-and- metadata** = the title/description/tags metadata (points HERE for the thumbnail) · **nano-banana**/**ideogram** = generate the raster this briefs · **tiktok-video-publishing** = the TikTok `cover` field · **hook-writer**/ **caption-writer** = text lines. ## Where this connects Reads first: **brand-profile** + the **video/title** (**youtube-publishing-and-metadata**/**youtube-long- form**/**youtube-shorts**). Pairs with: **nano-banana**/**ideogram** (render the brief), **hook-writer** (the thumbnail words), **youtube-publishing-and-metadata** (title as a unit with the thumbnail), **experimentation** (A/B discipline), **competitor-analysis** (preview against ranking thumbnails). Hands to: a **human** to render + **upload the YouTube thumbnail natively** + run **Test & Compare**; **analytics-and-reporting** reads CTR/watch-time back. The raster, native upload, and native A/B test stay outside WoopSocial. ## Definition of done A one-sentence promise written and the thumbnail designed as a unit with the title (show vs tell, no repetition); one focal point with minimal background; a genuine face + gaze where it adds context (or none); high contrast (complementary, not YouTube red/white, 60-30-10, passes the B&W test) legible in light and dark; 0–4 bold words with an outline that don't duplicate the title; mobile-legible at ~168×94 with safe zones and 1280×720 specs; an A/B plan of 3 meaningfully-different concepts changing one variable, scored natively by watch-time share; the raster left to an image tool/designer and the YouTube upload + test left to native Studio; no clickbait, no unconsented faces, no fabricated CTR; correctly distinguished from youtube-publishing-and-metadata and nano-banana/ideogram.
© social-media-skills · MIT · social-media-skills/skills