S2 Docs MCP Server

A Model Context Protocol server that provides AI tools with access to Spectrum 2 documentation. Part of the Spectrum Design Data monorepo.

Features

• Search components by name, category, or content
• Get component docs with full formatting
• Statistics on scraped documentation
• Find by use case (e.g., "form input", "navigation")
• List all components by category

Installation

cd ~/Spectrum/spectrum-design-data/tools/s2-docs-mcp
pnpm install

Usage

Start the MCP Server

pnpm start

Or add to your .cursor/mcp.json:

{
  "mcpServers": {
    "s2-docs": {
      "command": "node",
      "args": [
        "/Users/YOUR_USERNAME/Spectrum/spectrum-design-data/tools/s2-docs-mcp/src/cli.js"
      ]
    }
  }
}

Batch Scraping

List all components:

pnpm run scrape list

Parse a snapshot file:

pnpm run scrape parse <snapshot-file> <category> <slug>

Example:

pnpm run scrape parse ~/.cursor/browser-logs/snapshot.log actions button

Generate component index (scans docs/s2-docs/):

pnpm run scrape index

Check scraping status:

pnpm run scrape report

Available MCP Tools

list-s2-components

List all available Spectrum 2 components, optionally filtered by category.

Parameters:

• category (optional): Filter by category (actions, containers, feedback, inputs, navigation, status)

get-s2-component

Get full documentation for a specific component.

Parameters:

• name (required): Component name or slug
• category (optional): Component category for faster lookup

search-s2-docs

Search documentation by component name or within content.

Parameters:

• query (required): Search query
• searchContent (optional, default: false): Search within component content

get-s2-stats

Get statistics about scraped documentation coverage.

find-s2-component-by-use-case

Find components by use case or design pattern.

Parameters:

• useCase (required): Use case description (e.g., "form input", "navigation")

Scraping Workflow

1. Use Cursor's scrape-s2-docs skill to scrape components into docs/s2-docs/
2. Regenerate index after adding or updating docs:

   pnpm run scrape index

3. Check status:

   pnpm run scrape report

4. Start the MCP server to query docs

Data Location

Documentation is read from:

spectrum-design-data/docs/s2-docs/
├── components/
├── designing/
├── fundamentals/
├── developing/
├── support/
└── index.json

See docs/s2-docs/README.md for maintenance and transform scripts.

Using with Cursor

• MCP (this server) – Add the s2-docs MCP server to .cursor/mcp.json (see Usage above). The AI can then use the tools to list, search, and fetch component docs on demand.
• @Files & Folders – In chat, reference the docs/s2-docs folder (or a subfolder like docs/s2-docs/components) so the AI gets the markdown files as context.
• @Docs – If the S2 documentation is published at a URL, add it in Cursor via @Docs → Add new doc so Cursor indexes it. See Cursor’s @Docs documentation.

Integration with spectrum-design-data-mcp

This server complements the existing spectrum-design-data-mcp tool in this monorepo:

• spectrum-design-data-mcp: Design tokens and component schemas
• s2-docs-mcp: S2 component documentation and design guidelines

Both can run simultaneously in Cursor.

License

Apache-2.0 © Adobe