MCP Server
Playroom docs are exposed as an MCP (Model Context Protocol) server so AI assistants (Cursor, Claude Desktop, Replit, etc.) can read documentation programmatically instead of relying on pasted context.
Endpoint
https://docs.joinplayroom.com/api/mcpUse this URL in your MCP client configuration.
Connecting
Cursor
Add to Cursor Settings → MCP → Edit Config (or .cursor/mcp.json):
{
"mcpServers": {
"playroom-docs": {
"url": "https://docs.joinplayroom.com/api/mcp"
}
}
}Claude Desktop
Edit your Claude Desktop config (e.g. ~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
"mcpServers": {
"playroom-docs": {
"url": "https://docs.joinplayroom.com/api/mcp"
}
}
}stdio-only clients
If your client only supports stdio, use mcp-remote as a proxy:
{
"mcpServers": {
"playroom-docs": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://docs.joinplayroom.com/api/mcp"]
}
}
}Tools
The server exposes these tools:
| Tool | Description |
|---|---|
| docs_toc | Markdown table of contents for the whole site (all pages with links and descriptions). Use first to discover pages. |
| docs_page | Single page content by slug (e.g. examples/live-cursors, api-reference). Use after docs_toc or docs_search to get the right slug. |
| docs_full | Entire documentation as plain text (all pages with frontmatter). Use when full context is needed. |
| docs_toc_section | Table of contents for one section only (e.g. api-reference, examples). |
| docs_full_section | Full content for one section (all pages under that path). |
| docs_search | Search by keyword (titles, descriptions, content). Returns matching slugs and titles for use with docs_page. |
Slugs are paths without a leading slash: api-reference, examples/live-cursors, concepts/ai-features.
Related
- AI features — Overview of Playroom’s AI support (MCP, npm skill).
- llms.txt — Same content as docs_toc in a single text file for manual or scripted use.