Public MCP Server

Aira exposes these docs over a public, read-only Model Context Protocol (MCP) endpoint. Any AI client that speaks MCP can connect to it and answer questions about Aira straight from the source of truth — no copy-pasting, no scraping.

Endpoint

https://lxaira.com/api/mcp
  • Protocol: MCP JSON-RPC 2.0 over HTTP (POST)
  • Auth: none, it's public
  • CORS: open

Tools exposed

  • list_docs — returns every documentation page with its slug, title, and one-line summary.
  • get_doc({ slug }) — returns the full markdown for a single page by slug.

The same pages are also exposed as MCP resources under docs://aichat/<slug>, so clients that prefer resources/list + resources/read work too.

Claude Desktop

Add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "aira-docs": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://lxaira.com/api/mcp"]
    }
  }
}

Restart Claude Desktop. The list_docs and get_doc tools will show up in the MCP panel.

Cursor

Add to your ~/.cursor/mcp.json (or workspace .cursor/mcp.json):

{
  "mcpServers": {
    "aira-docs": {
      "url": "https://lxaira.com/api/mcp"
    }
  }
}

Test it from the command line

curl -s -X POST https://lxaira.com/api/mcp \
  -H 'content-type: application/json' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

curl -s -X POST https://lxaira.com/api/mcp \
  -H 'content-type: application/json' \
  -d '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"list_docs","arguments":{}}}'

curl -s -X POST https://lxaira.com/api/mcp \
  -H 'content-type: application/json' \
  -d '{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"get_doc","arguments":{"slug":"widget-embed"}}}'

Why an MCP for docs?

Because "read the docs yourself" is a better AI experience than "let me tell you what I remember about the docs". When your AI assistant can pull the current, authoritative page on demand, you get less hallucination and fewer outdated recommendations.

Public MCP Server — Docs — Aira