difydocmcpserver

# Dify Doc MCP Server A MCP (Model Context Protocol) Server for integrating Dify knowledge base, allowing AI assistants to directly access and retrieve content from Dify knowledge base. ## Technical Selection - **Language**: TypeScript - **Runtime Environment**: Node.js v18+ - **MCP Framework**: @modelcontextprotocol/sdk - **HTTP Client**: axios - **Data Validation**: zod - **Transport Method**: StdioServerTransport (standard input/output) ## Main Architecture This project implements three core tools: 1. **dify_retrieve_knowledge** - Knowledge Base Retrieval 2. **dify_list_datasets** - Get Knowledge Base List 3. **dify_get_dataset_detail** - Get Knowledge Base Details Through MCP protocol integration with Dify REST API, it supports multiple retrieval methods such as semantic search, keyword search, full-text search, and hybrid search. ## Installation ### Global Installation ```bash npm install -g dify-doc-mcp-server ``` ### Or use npx (no installation required) ```bash npx dify-doc-mcp-server ``` ## Quick Start ### Method 1: Using npx (recommended) ```bash # Run directly without installation DIFY_BASE_URL="http://your-dify-host:8031" DIFY_API_KEY="your-api-key" npx dify-doc-mcp-server ``` ### Method 2: Local Development ```bash # 1. Clone the project git clone <repository-url> cd difydocmcpserver # 2. Install dependencies npm install # 3. Configure environment variables export DIFY_BASE_URL="http://localhost:8031" export DIFY_API_KEY="your-dify-api-key" # 4. Build and run npm run build npm start ``` ## Available Tools ### dify_list_datasets Get Dify knowledge base list - `keyword` (optional): Search keyword - `limit` (optional): Return count (1-100, default 20) ### dify_get_dataset_detail Get Dify knowledge base details - `dataset_id` (required): Knowledge base ID ### dify_retrieve_knowledge Retrieve relevant content from Dify knowledge base - `dataset_id` (required): Knowledge base ID - `query` (required): Retrieval query content - `search_method` (optional): Retrieval method (semantic_search/keyword_search/full_text_search/hybrid_search) - `top_k` (optional): Return result count (1-20, default 3) - `score_threshold` (optional): Similarity threshold (0-1, default 0) ## MCP Client Configuration ```json { "mcpServers": { "dify-doc": { "command": "npx", "args": ["dify-doc-mcp-server"], "env": { "DIFY_BASE_URL": "http://your-dify-host:8031", "DIFY_API_KEY": "your-dify-api-key" } } } } ``` ## API Key Acquisition 1. Log in to Dify console 2. Go to "Settings" → "Knowledge Base" → "API" → "API Key" 3. Create a new API key 4. Copy the key and set it to environment variables ## Testing ```bash # Set environment variables and run test DIFY_BASE_URL="http://your-dify-host:8031" DIFY_API_KEY="your-api-key" DATASET="your-dataset-name" QUERY="引导" node test-mcp.js ``` The test script will verify: - MCP Server connection - Tool list acquisition - Knowledge base list, details, and retrieval functionality ## Related Links - [Dify Official Documentation](https://docs.dify.ai/) - [MCP Protocol Specification](https://modelcontextprotocol.io/) - [MCP TypeScript SDK](https://github.com/modelcontextprotocol/typescript-sdk)