Brand covers voice, visual identity, messaging frameworks, asset management and consistency reviews. Its source of truth is a brand guidelines file in the project, and four Node scripts turn that file into prompt context, design tokens, asset checks and color comparisons.
Reference documents cover voice, messaging, logo usage, color palettes, typography, asset organization and approval checklists, and a starter template gives a new brand a complete guidelines file to fill in.
What You Get
- Four Node scripts for brand context injection, token sync, asset validation and color extraction
- Ten reference documents on voice, visual identity, messaging, logos, colors, typography, assets and approvals
- A starter brand guidelines template for new brands
- A sync flow from docs/brand-guidelines.md to assets/design-tokens.json and design-tokens.css
- An update subcommand for revising the brand and syncing it to design systems
Common Use Cases
- Writing a first brand guidelines file for a new product
- Keeping marketing copy in a consistent voice by injecting brand context into prompts
- Auditing logos and images for naming, size and format rules
- Turning brand colors and fonts into CSS variables without hand copying
What Makes It Unique
Many brand skills are style advice only. This one connects the guidelines file to code: a single sync produces token JSON and CSS, and the sync stops when it finds an existing token setup, CSS custom properties or a Tailwind theme, so it does not overwrite a design system the project already has. It also serves as the first stage of the repository’s wider design workflow, feeding the design-system and ui-styling skills.
Under The Hood
The model reads SKILL.md and the references it needs, then runs Node scripts from the skill folder with the project as the working directory. The scripts read the brand guidelines and project files, and the sync script writes token JSON and CSS into the assets folder and calls the design-system token generator as a child process. It refuses to proceed when it detects existing token or theme files unless force is passed. No network calls are made.