Knowledge MCP Server

Example Use Cases

• 1

  Consult curated, domain-specific knowledge bases during reasoning to produce more accurate, context-aware answers.

• 2

  An MCP-accessible retrieval layer combining vector search and a knowledge graph for entity- and relationship-aware queries.

• 3

  Manage and update local knowledge stores (ingest documents, configure parsing/search behavior, and tune response prompts) via a CLI.

Description

knowledge-mcp lets you create, manage, and query domain-specific knowledge bases and expose them to MCP-compatible AI clients via a FastMCP server. It uses LightRAG to ingest documents, build embeddings and a knowledge graph, and supports multiple retrieval modes, text-only parsing, and configurable response prompts. The project provides a CLI for KB lifecycle management and expects local configuration and optional LLM/embedding API keys (e.g., OpenAI) for processing and querying. This enables agents to consult authoritative, context-aware information during their chain-of-thought.