MCP integration

Model Context Protocol — client mode. Connect to external MCP servers and let the chatbot use their tools via function calling.

What it does

  • Connect to external MCP servers (streamable HTTP or SSE transport) from the Settings → MCP tab.
  • Discover & cache each server’s tool list (click Discover).
  • Discovered tools are normalized to an OpenAI function-calling schema and offered to the LLM.
  • The chatbot runs a bounded tool-call loop (max 5 iterations), invoking tools/call on the MCP server and feeding the results back into the conversation.
  • Tool invocations surface as a 🔧 Using tool: X badge in the chat stream.

Protocol details

The MCP client implements JSON-RPC 2.0 over streamable HTTP:

  • initialize
  • notifications/initialized
  • tools/list
  • tools/call

Session-id capture and SSE-aware response extraction are handled automatically, so both streamable-HTTP and SSE transports work.

Adding a server

  1. Go to ItihRag → Settings → MCP.
  2. Click Add server and enter the server’s URL (and any auth required).
  3. Click Discover to fetch and cache the tool list.
  4. Enable the server. Its tools are now available to the chatbot.

Admin REST endpoints

Method Path Purpose
GET / POST /mcp/servers List / add MCP servers
POST / DELETE /mcp/servers/<id> Update / delete a server
POST /mcp/servers/<id>/discover Refresh a server’s tool list

All MCP endpoints require the manage_options capability.