brand-to-images
Skill17 starsChanged 8 months agoMITView on GitHub
Transforms brand guidelines into optimized image prompts for Manus + Nano Banana, covering hero images, backgrounds, icons, OG images, and more.
In this repository20 other files it gives its agents.
- aios-presentation-standardSkill
- audit-existing-siteSkill
- brand-voiceSkill
- code-auditorSkill
- content-auditorSkill
- copy-editorSkill
- deploy-to-vercelSkill
- form-wiringSkill
- Getting Started with VibeCodingSkill
- link-checkerSkill
- performance-optimizerSkill
- qa-testerSkill
- security-auditorSkill
- seo-aeo-auditorSkill
- site-customizerSkill
- site-generatorSkill
- site-interviewerSkill
- site-updaterSkill
- ux-reviewerSkill
- webmasterSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
15 rules. About 0 are common, 13 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈13
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 46
Before generating prompts, understand these limitations:
- Distinctive: line 50
Should avoid generating specific people (use generic archetypes instead)
- Very specific: line 62
If loading from site-brief.json:
- Distinctive: line 220
Be visual, not abstract: Instead of "convey trust," say "people shaking hands in a bright, open office."
- Distinctive: line 221
Avoid forbidden content: No embedded text, no branded logos, no trademarked products.
- Distinctive: line 223
Include context clues: If the image is for a financial services company, describe "modern office setting" or "professional workspace."
- Distinctive: line 229
Contradictory direction: Don't ask for "bold yet minimal" or "energetic and calm." Choose one and embrace it.
- Distinctive: line 236
If a generated image isn't quite right:
- Distinctive: line 237
Adjust the prompt,: not the generated image
- Distinctive: line 248
Features: Use icons or illustrated diagrams (minimal text)
- Very specific: line 281
After completing site-interviewer and having a site-brief.json
- Distinctive: line 282
When you want to establish visual consistency across your website
- Distinctive: line 283
Before you start designing or building HTML
- Distinctive: line 284
When you need to brief a designer or developer on visual direction
- Distinctive: line 285
When you want to generate hero images, backgrounds, and supporting visuals quickly
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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.
Files
Brand-to-Images Skill
This skill takes your brand guidelines and transforms them into image prompts optimized for Manus (image generation orchestration) and Nano Banana (Claude's visual generation model). The result is a collection of production-ready prompts that you can feed directly into Manus to generate cohesive, on-brand images.
How This Works
Input: Either a site-brief.json from the site-interviewer skill, or brand guidelines you provide directly.
Process:
- Extract brand elements (colors, tone, mood, industry)
- Map your website pages to image needs (hero, sections, team, etc.)
- Generate detailed, visual prompts optimized for Nano Banana
- Organize by page and image type for easy reference
- Include usage guidance for each image type
Output: A image-prompts.md file with prompts organized by page, plus guidance on how to use Manus to execute them.
Key Concepts
Nano Banana Best Practices
Nano Banana (Claude's image generation model) works best when:
- Prompts are visual and specific (describe what you see, not abstract concepts)
- Prompts avoid text within images (no logos, watermarks, or words overlaid)
- Prompts include style/mood descriptors (not just "a photo of X")
- Prompts are concise but vivid (200 words is the sweet spot)
- Prompts use photographic or illustrative language depending on the image type
What You Cannot Do with Images
Before generating prompts, understand these limitations:
- Cannot generate text/logos inside images (must be added in design)
- Cannot generate faces that are too recognizable (for legal reasons)
- Cannot generate trademarked products or brand logos
- Should avoid generating specific people (use generic archetypes instead)
See references/nano-banana-prompt-guide.md for complete best practices.
Skill Workflow
Step 1: Load Brand Data
You can provide:
- Option A: A
site-brief.jsonfile (from site-interviewer) - Option B: Brand guidelines directly in this conversation
If loading from site-brief.json:
Please load /path/to/site-brief.json and extract:
- Brand colors and palette
- Brand tone/mood
- Website pages needed
- Industry/business type
- Target audience descriptionIf providing directly:
Brand Name: [X]
Industry: [X]
Target Audience: [X]
Brand Colors: [primary], [secondary], [accent]
Brand Tone: [mood descriptors]
Website Pages: [home, about, services, contact, etc.]Step 2: Map Image Needs
Based on pages and industry, I'll identify what images you need:
- Home Page: Hero image, key section backgrounds
- About Page: Team photos (or illustrated team), office/workspace, mission visual
- Services/Products: Section headers, product showcases, benefit illustrations
- Contact Page: Office location image or contact illustration
- Blog: Featured images, category headers
- General: Social media OG images, favicon concept, footer imagery
Step 3: Generate Image Prompts
For each image, I'll create:
- Nano Banana Prompt — Optimized for image generation
- Image Specs — Size, format, usage context
- Mood/Style — Specific visual style (photographic, illustrated, etc.)
- Variation Notes — How to adjust for different contexts
Step 4: Output image-prompts.md
Organized like this:
# Image Prompts: [Business Name]
## Brand Summary
[Thumbnail of brand: colors, tone, visual style]
## Home Page
### Hero Image
**Nano Banana Prompt:**
[Detailed visual prompt]
**Specs:**
- Dimensions: 1920x1080px
- Format: JPG with optimized for web
- Alt text: [accessible description]
- Usage: Full-width above the fold
**Style Guide:**
- Mood: energetic, professional, approachable
- Color harmony: Should incorporate [primary color] naturally
- Photography style: Lifestyle/authentic (not stock photo vibes)
---
## About Page
### Team Illustration
[Similar structure]
---
Image Type Library
See references/image-types.md for detailed specifications for:
- Hero images (dimensions, safe zones, text overlays)
- Background images (patterns, opacity, text overlay compatibility)
- Section headers (aspect ratios, focal point guidance)
- Team/People images (diversity, approachability, business casual)
- Icons & illustrations (stroke weight, color usage, scale)
- OG/Social images (dimensions per platform, text integration)
- Favicon concepts (simplicity, scalability)
Prompt Template Examples
Here are template examples for different moods. Adapt these based on brand:
Professional & Trustworthy
A [photography style] image of [subject] in a modern, clean environment.
The composition conveys [tone]: [specific details].
Color palette emphasizes [primary color] and [secondary color], with natural lighting.
Sharp focus on [key element], with subtle [background detail].
Professional and approachable, not stiff or corporate.Playful & Creative
A vibrant [style] image of [subject] with energy and personality.
The composition is dynamic and engaging, showing [specific moment].
Color palette leans into [primary color] with bursts of [accent color].
The mood is friendly and approachable, inviting [audience] to connect.
Authentic and human, avoiding clichés.Minimal & Modern
A clean, minimalist [style] composition of [subject].
Plenty of negative space, with focus on essential elements.
Limited color palette: [primary], [secondary], and white/neutral.
Geometric or natural lines create visual interest without clutter.
Modern, sophisticated, and timeless.Bold & Ambitious
A striking, dynamic [style] image of [subject] with confidence and impact.
High contrast, bold use of [primary color] and [secondary color].
Composition draws the eye to [focal point] with dramatic lighting or angle.
The mood is forward-thinking and transformative.
Contemporary and eye-catching.How to Use These Prompts with Manus
Once you have image-prompts.md, the workflow is:
- Copy a prompt from the markdown file
- Open Manus (or invoke it via Claude Code)
- Paste the prompt and specify:
- Dimensions (if not in prompt)
- Any style variations (e.g., "more illustration, less photography")
- Number of variations to generate (usually 3)
- Manus routes to Nano Banana and generates images
- Download and review the generated images
- Integrate into your design (Figma, HTML, etc.)
Example Manus invocation:
Generate image using Nano Banana:
Prompt: [paste prompt from image-prompts.md]
Dimensions: 1920x1080
Variations: 3
Style override: "more illustration, less photography"Tips for Success
Writing Image Prompts
- Be visual, not abstract. Instead of "convey trust," say "people shaking hands in a bright, open office."
- Avoid forbidden content. No embedded text, no branded logos, no trademarked products.
- Use sensory language. "Warm, soft lighting" vs. "good lighting." "Rich blues and golds" vs. "nice colors."
- Include context clues. If the image is for a financial services company, describe "modern office setting" or "professional workspace."
- Specify the vibe. Are we aiming for authentic/real, stylized, illustrated, photographic? Say it.
Avoiding Common Mistakes
- Too many words. Nano Banana doesn't get better with 500-word prompts. Clear and vivid beats verbose.
- Contradictory direction. Don't ask for "bold yet minimal" or "energetic and calm." Choose one and embrace it.
- Overthinking the brand colors. Colors should appear naturally, not forced. A hero image about cooking doesn't need to be aggressively branded.
- Vague subjects. "A team" is worse than "Five diverse professionals in business casual, laughing in a sunlit office."
- Trying to embed text. Nano Banana can't add readable text. Never ask for "a poster with the words 'Contact Us.'"
Revising Generated Images
If a generated image isn't quite right:
- Adjust the prompt, not the generated image
- Common tweaks: "more colorful," "less saturated," "add more people," "emphasize [element]"
- Generate 3 variations and compare
- Save the best one; regenerate if none fit
Common Image Scenarios by Business Type
SaaS Company
Hero: Dashboard or person using software, with confident posture and bright, tech-forward environment About: Team collaboration, diverse group in modern office Features: Use icons or illustrated diagrams (minimal text) Social OG: Bold typography area + illustrative background
Ecommerce Brand
Hero: Lifestyle shot of product in context (e.g., shoes on someone running, not just shoe photos) Category Pages: Product closeups with styling context About: Founder or team in the space where products are made Social OG: Product hero + lifestyle context
Professional Services (Law, Accounting, Consulting)
Hero: Confident professional in modern office setting, or abstract visual (graphs, handshake) that suggests partnership About: Team photos, office space, client collaboration Services: Icons and illustrative diagrams for each service Social OG: Minimal, professional layout with accent color
E-learning / Education
Hero: Engaged learner, diverse group in learning moment, or abstract visual of growth/knowledge Course Pages: Section illustrations showing learning outcomes Testimonials: Real or illustrated faces with warm, approachable mood Social OG: Inspiring, motivational visual
Creative Services (Design, Marketing, Branding)
Hero: Creative work in progress, inspired workspace, or abstract visual of creativity Portfolio: Showcase work samples (if possible) or abstract representations About: Founder/team in creative workspace Social OG: Bold, creative, eye-catching composition
When to Use This Skill
- After completing site-interviewer and having a
site-brief.json - When you want to establish visual consistency across your website
- Before you start designing or building HTML
- When you need to brief a designer or developer on visual direction
- When you want to generate hero images, backgrounds, and supporting visuals quickly
What Comes Next
Once you have image-prompts.md:
- Run prompts through Manus to generate actual images
- Review and iterate — adjust prompts if needed, regenerate
- Download images for use in design and development
- Integrate into site-builder — feed generated images into your HTML/CSS
- Optimize for web — compress, lazy-load, ensure fast performance
Example: From Brand to Images
Input (from site-brief.json):
- Business: "Bloom Design" (branding agency)
- Tone: "professional but approachable, creative, trustworthy"
- Target audience: "Small business founders, busy, non-designers"
- Colors: Navy, gold, white
- Pages: Home, About, Services, Contact, Blog
Output (image-prompts.md excerpt):
Home Hero Image
Prompt: "A modern office scene with a diverse group of three professionals collaborating at a wooden table. They're reviewing design mockups and mood boards, engaged and focused. Warm, natural lighting from a large window. Minimalist decor with navy and gold accents. The mood is creative yet professional, approachable and collaborative. Photography style: lifestyle, authentic, not staged."
Specs:
- Size: 1920x1080px
- Alt: "Team reviewing design work together"
- Usage: Full-width hero above fold
This approach ensures every image reinforces your brand, speaks to your audience, and can be generated quickly using Manus and Nano Banana.
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
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.

