open-docs
An open-source MCP implementation providing document management functionality
claude mcp add --transport stdio askme765cs-open-docs-mcp npx -y open-docs-mcp --docsDir /path/to/docs
How to use
open-docs-mcp is an MCP server that provides a document management experience with crawling, indexing, and searching documents. It exposes MCP tools to enable or disable specific docs, crawl enabled docs, build a search index, and perform full-text searches across indexed content. To run it, you typically start the server via npx and point it at a directory containing your docs with the --docsDir flag. Once running, you can interact with the server through the MCP protocol using the available tools and prompts to manage your document sources and queries. This setup is compatible with Cursor-like Docs capabilities, enabling document indexing, access, and management via MCP prompts and resources. The included tools are enable_doc, disable_doc, crawl_docs, build_index, search_docs, list_enabled_docs, and list_all_docs, which you can invoke as part of your MCP workflow to control and query your docs.
How to install
Prerequisites:
- Node.js >= 16.x
- npm (comes with Node.js) or pnpm/yarn if preferred
Install and run using npx (no local install required):
- Ensure Node.js is installed
- Check: node -v and npm -v
- Run the MCP server with docs directory:
npx -y open-docs-mcp --docsDir ./docs
Optional: Install locally for offline usage or development
- Clone the repository (if applicable) or initialize a project
- Install dependencies
npm install - Start the server (if a local entry point is provided, e.g., via package.json scripts)
npm run start
Configuration for client integration (e.g., Claude Desktop):
- Use the following MCP server config to connect:
{ "mcpServers": { "open-docs-mcp": { "command": "npx", "args": [ "-y", "open-docs-mcp", "--docsDir", "/path/to/docs" ] } } }
Note: Replace /path/to/docs with your actual documentation directory.
Additional notes
Tips and common issues:
- Ensure your docsDir path is accessible by the process running the MCP server.
- If you modify the docs sources, you may need to run crawl_docs to refresh the index before searching.
- The server exposes tools like enable_doc and disable_doc to manage which docs are crawled; use list_enabled_docs to confirm current sources.
- If you require environment configuration (e.g., different docs sources or credentials), consider extending the mcpServers config with an env field as needed by your runtime environment.
Related MCP Servers
any-chat-completions
MCP Server for using any LLM as a Tool
bitbucket
Bitbucket MCP - A Model Context Protocol (MCP) server for integrating with Bitbucket Cloud and Server APIs
time
⏰ Time MCP Server: Giving LLMs Time Awareness Capabilities
unity-editor
An MCP server and client for LLMs to interact with Unity Projects
website-publisher
AI Website builder and publisher MCP. Quickly publish and deploy your AI generated code as real website URL. Support html, css, js, python etc.
xgmem
Global Memory MCP server, that manage all projects data.