MCP Integration Overview
Knowledge Tree provides a Model Context Protocol (MCP) server that gives any MCP-compatible AI client read-only access to the knowledge graph. This means you can explore nodes, facts, dimensions, and relationships directly from tools like Claude Desktop.
What is MCP?
The Model Context Protocol is an open standard that lets AI assistants connect to external data sources and tools. Instead of copying data into prompts, MCP provides structured, on-demand access to live data.
What the Knowledge Tree MCP server provides
The MCP server exposes 8 tools for navigating the knowledge graph:
| Tool | Purpose |
|---|---|
search_graph | Find nodes by text search |
get_node | Load node details, definition, counts |
get_dimensions | Load multi-model analyses (paginated) |
get_edges | Load relationships, sorted by evidence strength |
get_facts | Load facts grouped by source, with powerful filtering |
get_fact_sources | Load deduplicated source list for provenance |
search_facts | Search the global fact pool |
get_node_paths | Find shortest paths between two nodes |
All tools are read-only — they query the graph but never modify it.
Authentication
The MCP server uses Bearer token authentication. You generate an API token from the Research App's profile page, then include it in your MCP client configuration.