Knowledge

Exposes domain-specific knowledge bases via MCP for proactive AI access and reasoning.

python

33

GitHub Stars

python

Language

5 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "olafgeibig-knowledge-mcp": {
      "command": "uvx",
      "args": [
        "knowledge-mcp",
        "--config",
        "<absolute-path-to-your-config.yaml>",
        "mcp"
      ],
      "env": {
        "OPENAI_API_KEY": "YOUR_OPENAI_API_KEY"
      }
    }
  }
}

You run a focused MCP server that connects AI agents to specialized knowledge bases. It lets you manage, query, and retrieve domain-specific information through a FastMCP-compatible interface, so agents can access contextual knowledge during reasoning and produce more accurate, context-aware responses.

How to use

You interact with the server using an MCP client that understands the Model Context Protocol. Start the MCP server, point your client to the provided MCP endpoint, and then query your knowledge bases just like you would ask a question to a local assistant. The server processes documents, builds a knowledge graph, and returns relevant context to the client to inform responses.

How to install

Prerequisites: you need Python 3.12 and the uv tool installed.

Install the package in editable mode so you can develop and run from source.

# Install in editable mode (example)
uv pip install -e .

Run the MCP server via the CLI, supplying your main configuration path.

uvx knowledge-mcp --config /path/to/config.yaml shell

To allow an MCP client to connect, configure the MCP endpoints for the client. This example shows how to expose the server to a client via the MCP configuration structure.

{
  "mcpServers": {
    "knowledge_mcp": {
      "command": "uvx",
      "args": [
        "knowledge-mcp",
        "--config",
        "<absolute-path-to-your-config.yaml>",
        "mcp"
      ],
      "env": [
        {"name": "OPENAI_API_KEY", "value": "YOUR_OPENAI_API_KEY"}
      ]
    }
  }
}

Additional setup steps

Prepare the main configuration and environment file before starting. Copy the example configurations, then edit them to enable your API keys and adjust paths as needed.

Example commands to create and configure a knowledge base (described in the configuration steps) involve creating a base configuration, adding documents, and adjusting per-KB settings.

Available tools

knowledge-mcp CLI

A command-line interface to manage knowledge bases, including creating KBs, adding documents, configuring per-KB settings, querying, and inspecting storage.

LightRAG engine

Core knowledge base processing pipeline that ingests documents, chunks content, extracts entities/relations, builds a knowledge graph, and generates vector embeddings for retrieval.

MCP server interface

Exposes the search capability of knowledge bases via the FastMCP protocol so compatible AI agents can query directly.