GTC Design Tokens

Skill and template for the GTC tokens model. GTC is a design token framework that helps product teams and AI manage and scale tokens in large-scale UI systems. The model is built around three principles:

• Naming: how to name tokens so they stay clear as the system grows.
• Composition: how to group tokens so they're easy to manage and extend.
• Scope: how far a change travels, so a value updates globally, per theme, or per component, and never further than it needs to.

You can teach AI to use the model by installing the /gtc-tokens skill.

Using as a Claude Code skill

If you use Claude Code, clone this repo and copy the skill into your skills folder:

git clone https://github.com/bunind/gtc-tokens.git
cp -r gtc-tokens/skill ~/.claude/skills/gtc-tokens

Then in any Claude Code session, run /gtc-tokens. The skill lets you audit a token set against the GTC model, create tokens and components, sync with another .json file, or start a new project from the bundled template.

What's here

• skill/: the /gtc-tokens Claude Code skill. Copy it into ~/.claude/skills/gtc-tokens/ to use. It bundles its own copy of the template so :new works standalone, and that copy stays identical to template/.

• template/: the starter token set: global/, theme/, and component/, one JSON file per scale or component. Run python3 validate.py to check it against the GTC model.

• README.md: the GTC rulebook, written for AI agents. It covers the model, taxonomy, modes, and scale keys. AI reads it before adding or editing a token.

• .github/workflows/check.yml: CI for maintainers. On every push it confirms the skill's bundled copy of the template hasn't drifted from template/, and that both token sets still pass validate.py.

Skill commands

• :audit — review the token set against the GTC model.
• :create — add tokens or a component.
• :sync — sync tokens with another .json file.
• :new — start a new GTC project from the template.

License

MIT LICENSE