mcpserver.so LogoMCPFly
Submit
McpDocServer logo

McpDocServer

by ruan11223344

Development/DocumentationDocumentationCrawlerAPIFramework

McpDocServer is a development documentation server based on the MCP protocol, designed specifically for various development framework documentation. It offers features like multi-threaded document crawling, local document loading, keyword search, and document detail retrieval.

View on GitHub

Last updated: N/A

What is McpDocServer?

McpDocServer is a server that provides accurate documentation retrieval capabilities, especially useful for addressing the 'hallucination' problem in AI tools like Cursor when working with development frameworks. It retrieves the latest and most accurate API information directly from official documentation sources.

How to use McpDocServer?

  1. Install dependencies using npm install. 2. Configure document sources in config/doc-sources.js. 3. Run the crawler using npm run crawl. 4. Start the server using npm start. 5. Configure Cursor with the server's absolute path in mcp.json.

Key features of McpDocServer

  • Loads framework documentation data from local JSON files

  • Provides powerful document search functionality

  • Offers document detail query

  • Automatically identifies available document sources

  • Supports targeted queries for specific document sources

  • Crawls external documents and automatically converts them to a locally usable format

  • Supports reloading documents (triggered by searching for 'reload')

Use cases of McpDocServer

  • Improving the accuracy of AI code completion tools like Cursor

  • Providing precise API documentation for developers

  • Creating a centralized documentation repository for multiple frameworks

  • Offline access to documentation

  • Quickly searching and retrieving information from large documentation sets

FAQ from McpDocServer

How do I configure the document sources?

Create or modify the config/doc-sources.js file to define the URLs and patterns for the documentation websites you want to crawl.

How do I run the document crawler?

After configuring the document sources, run the command npm run crawl to start the crawler.

How do I start the MCP server?

Run the command npm start to start the server. It will automatically load the documentation files from the docs directory.

How do I configure Cursor to use this server?

Add a new MCP server configuration to your mcp.json file, specifying the absolute path to the server.js file.

How do I reload the documentation without restarting the server?

Use the search_docs tool with the query 'reload'. This will trigger a reload of the documentation files.