Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when the user needs a complete static image brief for a designer or photographer: visual objective, concept, mood, composition, overlay copy, colors and fonts read from the brand guideline, channel dimensions, and approval criteria. Trigger on 'image brief', 'visual brief', 'banner brief', 'thumbnail brief', 'static brief', 'brief for the designer', or 'how should this graphic look'.
.claude/skills/minhnv0807-42-image-brief-global/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 119% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 54% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 117% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 151% | 0% |
A good image brief means the designer finishes without a single clarifying call. Communicate the feeling you want, not just "make it look good". Colors, fonts, and logo usage come from the brand guideline (46-brand-guideline-global), never invented here.
Read .agents/product-marketing-context-global.md, the brand guideline, and the asset list (41-campaign-asset-list-global) if available. If information is missing, ask up to 4 questions:
50-asset-resize-global).Before briefing the concept, decide what this image must accomplish in the first three seconds of attention.
| Objective | Visual direction | |-----------|------------------| | Stop the scroll | Strong hook, high contrast, pattern interrupt | | Build trust | Clean, professional, visible social proof | | Drive the click | Prominent CTA, explicit offer, urgency if it is real |
Fill every section of the template below. Leave nothing blank.
Always include 2-3 specific reference images, each annotated with what to learn from it (color? composition? mood?) and what not to copy.
File name: image-brief-[asset-name]-[YYYYMMDD].md
markdown# Image Brief — [Asset name] ## Overview | Item | Detail | |------|--------| | Project / campaign | | | Image type | | | Channel / placement | | | Dimensions (px) + ratio | | | Visual objective | Stop the scroll / Build trust / Drive the click | | Export format | JPG / PNG / WebP | | Number of deliverables | | | Designer | | | Deadline | | | Max revision rounds | | ## Concept and mood - One-sentence idea: [what this image is about — readable in one pass] - Feeling to convey: [Professional / Warm / Playful / Urgent / Minimal / Bold] - Mood board references: - Ref 1: [link or file — learn: color? layout? photographic style?] - Ref 2: [...] - Ref 3: [...] — do not copy: [...] - Colors (from the brand guideline): - Primary: [primary color] - Secondary / accent: [secondary color] - Background: [white / solid / gradient / photographic] - Text: [dark text on light ground / light text on dark ground] - Avoid entirely: color [...], style [...], element [...] ## Composition - Main subject: [real person / product / abstract / type-led / illustration] - Background: [white / solid / lifestyle scene / studio / blur / texture] - Focal point: [centered / rule of thirds left or right / full bleed] - Clear space for text overlay: [yes/no — which corner, roughly which third] - Visual hierarchy: 1. [most important element — where the eye lands first] 2. [second element] 3. [third element — CTA / logo] ## Overlay copy (approved before it is set) | Element | Content | Font | Size | Color | Position | |---------|---------|------|------|-------|----------| | Headline | | [heading font] Bold | | | | | Subtext | | [body font] Regular | | | | | CTA button | | [heading font] Bold | | | | | Logo | [version: horizontal / stacked / icon only] | — | — | — | [per brand guideline] | Copy limits: headline max [X] characters, subtext max [X] characters, no more than three text colors in one image. ## Deliverables - [ ] Source file (share link, not a flattened export) - [ ] Export [format] — [exact dimensions] - [ ] Variants if needed: [light/dark version, alternate crop] - [ ] Remaining sizes per the asset list ## Review process - Draft 1 due: [date] — feedback within [X] hours - Max revision rounds: [X] — final approval by: [lead] - Final deadline: [date]
| Channel | Dimensions | Safe zone | Export | Max file size | |---------|-----------|-----------|--------|---------------| | Meta Feed (FB/IG) | 1080x1080 (1:1) or 1080x1350 (4:5) | 20px each edge | JPG, RGB | 30MB | | Meta Story / Reels | 1080x1920 (9:16) | 250px top / 250px bottom | JPG, RGB | 30MB | | Meta link/landscape | 1200x628 (1.91:1) | 40px each edge | JPG, RGB | 30MB | | TikTok cover | 1080x1920 (9:16) | 150px top / 300px bottom | JPG | 10MB | | LinkedIn single image | 1200x627 (1.91:1) | 40px each edge | JPG/PNG | 5MB | | Pinterest pin | 1000x1500 (2:3) | 60px each edge | JPG/PNG | 20MB | | YouTube thumbnail | 1280x720 (16:9) | 60px each edge, avoid bottom-right timestamp | JPG/PNG | 2MB | | Email header | 600x200 | 10px each edge | JPG | 1MB | | Website banner | 1440x500 | 100px each edge | JPG/PNG/WebP | 500KB |
> Full size matrix, crop behavior, and resize strategy: read 50-asset-resize-global.
Image brief — Meta Feed 1080x1080 — TOFU, Q3 course launch
Type: single-image ad | Channel: Meta Feed | Objective: stop the scroll within 1.5 seconds
Feeling: bold, mild urgency, professional but not corporate
Subject: type-led, gradient background from [primary color] to [secondary color]
Headline: "[core message — max 60 characters]"
Subtext: "[product name + date window]"
CTA: "Enroll now"
Ref: [link to a bold type-led example — learn contrast and typography, do not copy the palette]
Deadline: [date] | Max revisions: 241-campaign-asset-list-global: where this image sits in the asset list and its deadline.46-brand-guideline-global: source of colors, fonts, and logo rules — read before writing the brief.45-design-tool-brief-global: add step-by-step production direction when the designer works in Canva, Figma, or Adobe Express.43-carousel-brief-global: use instead when the deliverable is a slide sequence rather than a single frame.47-design-review-global: score the result once the designer submits.50-asset-resize-global: adaptation rules when the same master runs at several sizes.30-design-master-global: when the image should be AI-generated instead of designed by hand.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | fail→pass | 17,933 | 14,370 | -20% | 1 | 1 | 0% | 2,223 | 4,859 | +119% | 0 | 0 | — |
case-06 | pass→pass | 13,324 | 12,528 | -6% | 1 | 1 | 0% | 1,409 | 3,469 | +146% | 0 | 0 | — |
case-01 | fail→pass | 19,052 | 20,974 | +10% | 1 | 1 | 0% | 3,383 | 5,193 | +54% | 0 | 0 | — |
case-02 | fail→fail | 28,363 | 19,633 | -31% | 1 | 1 | 0% | 5,195 | 5,757 | +11% | 0 | 0 | — |
case-03 | fail→pass | 27,836 | 20,963 | -25% | 1 | 1 | 0% | 3,833 | 4,816 | +26% | 0 | 0 | — |
case-04 | fail→pass | 17,559 | 19,555 | +11% | 1 | 1 | 0% | 2,225 | 4,832 | +117% | 0 | 0 | — |
case-07 | fail→fail | 12,823 | 19,864 | +55% | 1 | 1 | 0% | 2,343 | 5,541 | +136% | 0 | 0 | — |
case-08 | fail→pass | 7,667 | 5,877 | -23% | 1 | 1 | 0% | 1,261 | 3,169 | +151% | 0 | 0 | — |
case-09 | pass→pass | 19,501 | 8,209 | -58% | 1 | 1 | 0% | 2,259 | 3,496 | +55% | 0 | 0 | — |
case-10 | pass→pass | 12,077 | 12,790 | +6% | 1 | 1 | 0% | 1,883 | 4,394 | +133% | 0 | 0 | — |
case-11 | pass→pass | 13,073 | 11,391 | -13% | 1 | 1 | 0% | 2,054 | 4,220 | +105% | 0 | 0 | — |
case-12 | pass→pass | 11,391 | 7,269 | -36% | 1 | 1 | 0% | 1,861 | 3,337 | +79% | 0 | 0 | — |
case-13 | fail→pass | 9,372 | 5,901 | -37% | 1 | 1 | 0% | 1,590 | 3,107 | +95% | 0 | 0 | — |
case-14 | fail→pass | 12,879 | 6,815 | -47% | 1 | 1 | 0% | 2,147 | 3,259 | +52% | 0 | 0 | — |
case-15 | fail→pass | 9,222 | 4,568 | -50% | 1 | 1 | 0% | 1,536 | 2,953 | +92% | 0 | 0 | — |
case-16 | fail→pass | 17,023 | 15,382 | -10% | 1 | 1 | 0% | 2,756 | 4,728 | +72% | 0 | 0 | — |
case-17 | fail→fail | 11,768 | 11,074 | -6% | 1 | 1 | 0% | 1,967 | 3,923 | +99% | 0 | 0 | — |
case-18 | fail→pass | 7,863 | 2,569 | -67% | 1 | 1 | 0% | 1,391 | 2,667 | +92% | 0 | 0 | — |
case-19 | pass→pass | 8,269 | 2,988 | -64% | 1 | 1 | 0% | 1,302 | 2,727 | +109% | 0 | 0 | — |
case-20 | fail→pass | 8,598 | 2,901 | -66% | 1 | 1 | 0% | 1,390 | 2,692 | +94% | 0 | 0 | — |
case-21 | fail→fail | 6,583 | 8,689 | +32% | 1 | 1 | 0% | 1,061 | 3,606 | +240% | 0 | 0 | — |
case-22 | pass→pass | 16,068 | 28,020 | +74% | 1 | 1 | 0% | 2,923 | 7,309 | +150% | 0 | 0 | — |
case-23 | pass→pass | 20,795 | 18,525 | -11% | 1 | 1 | 0% | 3,854 | 5,511 | +43% | 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. 23 cases were attempted. The headline lift of +48 percentage points is the difference between those two pass rates over the 23 comparable cases. 2 cases got worse with the skill loaded, and they are 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.