Use Crystallize

Crystallize AI

Everything you need to use Crystallize with AI agents — skills, MCP server, and documentation.

Structure

This repo contains three sub-projects:

Project	Path	Description
Skills	use-crystallize/skills/	Markdown-based skill modules for AI agents (query, mutation, content-model, pricing, permissions, etc.)
MCP Server	use-crystallize/mcp-servers/crystallize/	Cloudflare Workers MCP server providing authenticated access to Crystallize APIs
Docs	docs/	Astro Starlight documentation site deployed to crystallizeapi.github.io/ai

Prerequisites

• Node.js (v20+)
• Bun

Getting Started

A root Makefile provides convenience targets for common tasks:

make install        # Install dependencies for all sub-projects
make codeclean      # Format (oxfmt) + type-check + lint
make tests          # Run the test suite
make serve-doc      # Start the documentation dev server
make serve-mcp      # Start the MCP dev server

Run make (or make list) to see all available targets.

Docs

cd docs
bun install
bun dev             # Start dev server
bun run build       # Production build
bun run preview     # Preview production build

MCP Server

cd use-crystallize/mcp-servers/crystallize
bun install
bun dev             # Vite dev server with HMR
bun build           # Production build
bun deploy          # Build + deploy to Cloudflare Workers
bun codeclean       # Lint + format (oxlint + oxfmt)
bun type-check      # TypeScript type checking

Skills

Skills are plain markdown files in use-crystallize/skills/ — no build step required. Each skill has a SKILL.md with YAML frontmatter and an optional references/ directory with supporting docs.

Available skills: query, mutation, content-model, js-api-client, permissions, pricing, information-architecture.

Using the Claude Plugin

The use-crystallize/ directory is a Claude plugin. The plugin configuration lives in use-crystallize/.claude-plugin/plugin.json and references:

• Skills from ./skills/
• MCP Server at https://mcp.crystallize.com/mcp

To connect to the MCP server, you need Crystallize access tokens configured in the headers:

X-Crystallize-Access-Token-Id: <your-token-id>
X-Crystallize-Access-Token-Secret: <your-token-secret>

Inspect the MCP Locally

First start the MCP dev server:

make serve-mcp

Then add and inspect it:

bunx add-mcp http://localhost:5173/mcp \
  --header "X-Crystallize-Access-Token-Id: xxx" \
  --header "X-Crystallize-Access-Token-Secret: xxx"

bunx @modelcontextprotocol/inspector

Contributing

Please read the contribution guidelines before submitting a pull request.

License

MIT — Copyright 2026 Crystallize.