ragflow-knowledge-mcp-server

What is ragflow-knowledge-mcp-server?

An MCP (Model Context Protocol) server that provides a knowledge base interface for RAGFlow, enabling AI assistants to search and retrieve knowledge from configured datasets. It is intended for developers integrating RAGFlow knowledge bases with MCP-compatible clients.

How to use ragflow-knowledge-mcp-server?

Install via pip (pip install ragflow-knowledge-mcp-server) or from source. Configure via a config.yaml file specifying the RAGFlow API base URL, API key, and dataset definitions. Run with ragflow-knowledge-mcp-server --config=/path/to/config.yaml, using Python or uv, or through Docker/Docker Compose. The server supports both stdio and SSE transports.

Key features of ragflow-knowledge-mcp-server

• Dynamic knowledge base searching tools configured per dataset.
• List knowledge bases (optional, disabled by default).
• Get information of a specific knowledge base (optional, disabled by default).
• Configurable via YAML file and environment variables.
• Supports both stdio and SSE transports.
• Compatible with RAGFlow versions 0.17.2 and 0.18.0.
• Logging via simp-logger with file and console output.

Use cases of ragflow-knowledge-mcp-server

• Integrate RAGFlow knowledge bases with MCP-compatible AI assistants.
• Enable dynamic search of specific knowledge datasets during AI interactions.
• List and inspect available knowledge bases for management purposes.
• Retrieve metadata about a particular knowledge base.

FAQ from ragflow-knowledge-mcp-server

What RAGFlow versions does this server support?

It supports RAGFlow versions 0.17.2 and 0.18.0.

How do I configure the server?

Configuration is done via a YAML file (default config.yaml) or environment variables. Required settings include default-base-url and default-api-key, plus dataset definitions.

Does the server support SSE transport?

Yes, SSE transport is supported (the default is stdio). Set transport: sse in the config and optionally configure sse-port.

Can I list all knowledge bases?

Yes, the list_knowledge_bases tool is available but disabled by default. Enable it by setting list-bases-enabled: true in the config.

What authentication is used?

The server uses an API key (api-key) to authenticate with the RAGFlow instance. This can be set globally or per dataset.