A polished open-source model-native skill that turns a brand name or short brief into:
- a sharp strategic brand system
- a standalone visual HTML brand board
- optional design tokens for implementation handoff
This repository is intentionally skill-first. The model does the work. The repository provides the rules, structure, examples, and quality bar.
Default artifacts:
output/brand-system.mdoutput/brand-board.html- optional
output/brand-tokens.css
The goal is not to generate a wall of branding text. The goal is to generate a presentation-ready concept system.
Most weak branding prompts produce:
- vague adjectives
- low-contrast direction options
- generic slogans
- long strategy text with no visual form
This skill is built to produce:
- three clearly different directions
- one confident recommendation
- cleaner strategic language
- stronger placeholder safety
- browser-ready visual boards with real composition
- mock concept frames that feel like an actual review deck
The repository is optimized for:
- direction contrast over superficial variation
- visual hierarchy over excessive copy
- role-based systems over random style choices
- placeholder honesty over fake realism
- outputs that can be shown, not just read
Copy this folder to:
~/.claude/skills/brand-launchpad/Copy it into your repository at:
.claude/skills/brand-launchpad/Invoke with:
/brand-launchpad your brief here
The skill works well with:
- a strong brand name only
- a one-line product idea
- category + audience + vibe
- a short brief with constraints
Examples:
/brand-launchpad Satya Kingdom Entertainment
/brand-launchpad premium AI study planner for students
/brand-launchpad bold creator-led media brand for Gen Z
/brand-launchpad minimal luxury salon for a women-led studio
- Trigger the skill.
- Give the brand name or short brief.
- Let the model write the output files.
- Open
output/brand-board.htmlin a browser. - Refine with follow-up prompts if needed.
brand-launchpad/
├── SKILL.md
├── README.md
├── AGENTS.md
├── CHANGELOG.md
├── CONTRIBUTING.md
├── CODE_OF_CONDUCT.md
├── SECURITY.md
├── LICENSE
├── .editorconfig
├── .gitignore
├── .github/
│ ├── ISSUE_TEMPLATE/
│ ├── PULL_REQUEST_TEMPLATE.md
│ └── workflows/
├── data/
│ ├── palettes.json
│ ├── font_pairs.json
│ └── style_directions.json
├── reference/
│ ├── decision-rules.md
│ ├── style-system.md
│ ├── output-template.md
│ ├── visual-output.md
│ └── html-board-checklist.md
├── templates/
│ └── brand-board-template.html
├── examples/
│ ├── ai-study-planner.board.html
│ ├── ai-study-planner.brand-system.md
│ ├── satya-kingdom-entertainment.brand-board.html
│ ├── satya-kingdom-entertainment.brand-system.md
│ ├── satya-kingdom-entertainment.brand-tokens.css
│ └── other example briefs
├── evals/
│ ├── README.md
│ └── cases.yaml
└── output/
└── .gitkeep
A strong board should feel like:
- a mini brand presentation
- a creative review board
- a browser-openable artifact with real hierarchy
The recommended direction should be obvious. The three directions should not feel like recolors of the same idea.
This repo is strict about invented proof.
If the model includes testimonials, media logos, names, or metrics that were not supplied by the user, they must be clearly marked as sample or concept content.
Use these to understand the expected standard:
examples/ai-study-planner.board.htmlexamples/satya-kingdom-entertainment.brand-board.htmlexamples/satya-kingdom-entertainment.brand-system.mdexamples/satya-kingdom-entertainment.brand-tokens.css
- Give the skill a category if the name alone is ambiguous.
- Give one or two constraints if you already know the vibe.
- Ask for refinement after the first board instead of overloading the initial prompt.
- Use the HTML board as the artifact you review and iterate on.
This repo is designed around Claude-style skill workflows, but the structure is also readable by other coding agents that can follow repository instructions and create files directly.
MIT