Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →When the user wants to plan, create, or repurpose visual content (images, infographics, social post images) across channels. Also use when the user mentions "content images," "social media images," "infographic," "visual content," "post image," "image specs," "visual repurposing," "content visuals," or "image for social post." For Pinterest, use pinterest-posts.
.claude/skills/mkurman-visual-content/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 58% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 51% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 80% | 0% |
-----------|-------------|-------| | Article / Blog | Hero image, in-article images, screenshots | See image-optimization for web (alt, WebP, LCP) | | Social post | Single image, carousel, or link preview | Platform-specific specs below | | Infographic | Primary format; data visualization | Repurpose to social (cropped), blog (full) | | Case study | Customer photo, results chart, logo | Repurpose to LinkedIn carousel, blog | | Product update | Screenshot, feature graphic | Changelog, email, social | | Email | Header image, inline graphics | Keep lightweight; many clients block images | | Landing page | Hero, trust badges, screenshots | See hero-generator, image-optimization |
| Context | Priority | Skill | |---------|----------|-------| | Website | Alt text, WebP, LCP, responsive, lazy loading | image-optimization | | Social posts | Platform dimensions, aspect ratio, file size | Platform skills (X, LinkedIn, Pinterest, etc.) | | OG / Twitter Cards | 1200×630, 1200×675 for link previews | open-graph, twitter-cards | | Email | Inline-friendly; avoid heavy images; alt for blocked | email-marketing |
| Platform | Post Image | Stories / Reels | Profile | Notes | |----------|------------|-----------------|---------|-------| | X (Twitter) | 1200×675 (16:9), 800×800 | — | 400×400 | See twitter-x-posts | | LinkedIn | 1200×627, 1200×1200; carousel up to 20 | — | 400×400 | See linkedin-posts; vertical preferred on mobile | | Pinterest | 1000×1500 (2:3) | — | 165×165 | Alt text ~25% more impressions; see pinterest-posts | | Instagram | 1080×1350 (4:5), 1080×1080 | 1080×1920 (9:16) | 320×320 | 4:5 outperforms square on feed | | Facebook | 1200×630, 1080×1080 | 1080×1920 | 320×320 | | | YouTube | Thumbnail 1280×720 | — | 800×800 | |
General: 1080px width works across most platforms; vertical (4:5, 9:16) outperforms square on mobile-first feeds. Keep critical elements (logo, text) in safe center—platforms may crop.
Principle: One core visual → multiple crops/formats → multiple channels.
| Core Visual | Adaptations | Channels | |-------------|-------------|----------| | Infographic | Full (blog), cropped sections (Instagram, LinkedIn carousel), square (X) | Blog, LinkedIn, Instagram, X | | Case study graphic | Hero (blog), single slide (LinkedIn), story (Instagram) | Blog, LinkedIn, Instagram | | Product screenshot | Hero (landing), post (X, LinkedIn), email header | Website, social, email | | Quote graphic | Square (X, LinkedIn), 4:5 (Instagram) | X, LinkedIn, Instagram |
Workflow: Design at largest needed size; export platform-specific crops. Use consistent colors, fonts, logo placement (see brand-visual-generator).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 20,832 | 25,520 | +23% | 1 | 1 | 0% | 3,662 | 5,804 | +58% | 0 | 0 | — |
case-02 | fail→pass | 33,468 | 23,713 | -29% | 1 | 1 | 0% | 5,362 | 5,499 | +3% | 0 | 0 | — |
case-03 | fail→pass | 32,054 | 22,361 | -30% | 1 | 1 | 0% | 5,187 | 5,158 | -1% | 0 | 0 | — |
case-04 | fail→pass | 15,003 | 13,844 | -8% | 1 | 1 | 0% | 2,539 | 3,823 | +51% | 0 | 0 | — |
case-09 | pass→pass | 16,981 | 18,352 | +8% | 1 | 1 | 0% | 2,600 | 4,118 | +58% | 0 | 0 | — |
case-05 | pass→pass | 10,308 | 7,122 | -31% | 1 | 1 | 0% | 1,570 | 2,620 | +67% | 0 | 0 | — |
case-06 | pass→pass | 10,550 | 11,494 | +9% | 1 | 1 | 0% | 1,666 | 3,376 | +103% | 0 | 0 | — |
case-07 | fail→fail | 12,519 | 12,692 | +1% | 1 | 1 | 0% | 1,969 | 3,363 | +71% | 0 | 0 | — |
case-08 | fail→pass | 11,867 | 12,297 | +4% | 1 | 1 | 0% | 2,036 | 3,674 | +80% | 0 | 0 | — |
case-10 | pass→pass | 7,957 | 7,545 | -5% | 1 | 1 | 0% | 1,354 | 2,644 | +95% | 0 | 0 | — |
case-11 | pass→pass | 13,743 | 16,253 | +18% | 1 | 1 | 0% | 1,999 | 3,797 | +90% | 0 | 0 | — |
case-12 | fail→pass | 16,834 | 19,059 | +13% | 1 | 1 | 0% | 2,433 | 4,354 | +79% | 0 | 0 | — |
case-13 | pass→pass | 15,653 | 15,952 | +2% | 1 | 1 | 0% | 2,459 | 4,130 | +68% | 0 | 0 | — |
case-14 | pass→pass | 4,729 | 7,144 | +51% | 1 | 1 | 0% | 819 | 2,603 | +218% | 0 | 0 | — |
case-15 | fail→fail | 5,051 | 6,706 | +33% | 1 | 1 | 0% | 811 | 2,586 | +219% | 0 | 0 | — |
case-16 | pass→pass | 15,753 | 18,645 | +18% | 1 | 1 | 0% | 2,415 | 4,608 | +91% | 0 | 0 | — |
case-17 | pass→fail | 13,899 | 11,357 | -18% | 1 | 1 | 0% | 2,347 | 3,477 | +48% | 0 | 0 | — |
case-18 | fail→pass | 13,678 | 12,298 | -10% | 1 | 1 | 0% | 2,117 | 3,462 | +64% | 0 | 0 | — |
case-19 | pass→pass | 17,431 | 16,239 | -7% | 1 | 1 | 0% | 2,685 | 4,103 | +53% | 0 | 0 | — |
case-20 | pass→pass | 6,710 | 9,899 | +48% | 1 | 1 | 0% | 1,276 | 3,405 | +167% | 0 | 0 | — |
case-21 | pass→pass | 6,333 | 7,737 | +22% | 1 | 1 | 0% | 1,219 | 3,027 | +148% | 0 | 0 | — |
case-22 | pass→pass | 7,717 | 8,076 | +5% | 1 | 1 | 0% | 1,259 | 2,659 | +111% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +27 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.