Skip to main content
Canton Network docs expose a public Model Context Protocol (MCP) server, so an AI agent can search and read this documentation directly instead of relying on training data or a generic web search. URL: https://docs.canton.network/mcp Transport: HTTP. Authentication: none — the server is public and advertised through https://docs.canton.network/.well-known/mcp.json.

Tools

The server exposes three tools to a connected MCP client: Use the exact names when setting tool permissions in your client.

Connecting a client

Claude Code

Claude Desktop

  1. Open Claude settings → Connectors
  2. Click Add custom connector
  3. Enter a name and https://docs.canton.network/mcp as the URL
  4. Click Add
  5. In a chat, use the attachments button to select the connector

Cursor

  1. Open the command palette (Cmd+Shift+P / Ctrl+Shift+P)
  2. Search Open MCP settings → Add custom MCP
  3. Add to mcp.json:
Any other MCP-compatible client (Goose, ChatGPT, and others) can connect the same way, pointing at the same URL.
Every page also exposes an MCP option in its contextual menu (next to Copy, Claude, ChatGPT), which copies a ready-to-use connection snippet for that page’s client.

Other agent-facing surfaces

Besides the MCP server, agents reach these docs through:
  • /llms.txt — a navigable index of the documentation, with large sections split into /_llms/*.txt sub-indexes
  • Per-page Markdown, at any page’s URL with .md appended
  • /skill.md — a project skill file summarizing Canton Network for coding agents