RAGFlow MCP Server

@wang-junjian

关于 RAGFlow MCP Server

暂无概览

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "ragflow-mcp-server": {
      "command": "uv",
      "args": [
        "sync"
      ]
    }
  }
}

工具

聊天助手的名称

数据集的 ID

聊天助手的会话 ID

提问内容

概览

What is RAGFlow MCP Server?

RAGFlow MCP Server is an MCP implementation that exposes RAGFlow’s API as tools for AI assistants. It enables listing datasets, creating chat assistants, and conducting conversations via MCP protocol. Aimed at developers integrating RAGFlow knowledge bases into Copilot, Continue, or Claude Desktop.

How to use RAGFlow MCP Server?

Install via uvx ragflow-mcp-server or run from source with uv. Configure with --api-key and --base-url arguments. Example JSON/YAML configs for Copilot, Continue, and Claude Desktop are provided in the README.

Key features of RAGFlow MCP Server?

Lists all datasets with ID and name
Creates a new chat assistant for a dataset
Conducts conversations with a chat assistant

Use cases of RAGFlow MCP Server?

Query company knowledge bases from an AI coding assistant
Dynamically create and chat with RAGFlow assistants in CI/CD pipelines
Build a chat interface on top of RAGFlow datasets using MCP-compatible clients
Automate retrieval-augmented generation workflows with external tools

FAQ from RAGFlow MCP Server

What does RAGFlow MCP Server do?

It acts as a bridge between MCP-compatible AI assistants and the RAGFlow API, allowing you to list datasets and chat with RAGFlow assistants without leaving your editor or chat app.

How do I configure RAGFlow MCP Server?

Pass --api-key (your RAGFlow API key) and --base-url (your RAGFlow server address) when launching. Example configs are available for Copilot, Continue, and Claude Desktop.

What runtime dependencies does it require?

Python and uv (or uvx for the published package). The server communicates over stdio via MCP.

Where does the data live?

All data resides in your RAGFlow instance. The MCP server only sends API requests to your configured --base-url.

How are requests transported and authenticated?

Transport is MCP over stdio. Authentication is via the RAGFlow API key provided in the --api-key argument.

常见问题

What does RAGFlow MCP Server do?

It acts as a bridge between MCP-compatible AI assistants and the RAGFlow API, allowing you to list datasets and chat with RAGFlow assistants without leaving your editor or chat app.

How do I configure RAGFlow MCP Server?

Pass `--api-key` (your RAGFlow API key) and `--base-url` (your RAGFlow server address) when launching. Example configs are available for Copilot, Continue, and Claude Desktop.