my-docs-mcp-server
0
my-docs-mcp-server is an MCP server enabling full-text indexation and search of Markdown files through MCP protocol. It features tools for efficient document search and retrieval, with a focus on high-precision handling of Japanese text.
my-docs-mcp-server
Overview
my-docs-mcp-server is a Model Context Protocol (MCP) server designed for full-text indexing and searching of Markdown files within a specified directory. Key features include:
- Offering full-text search as a tool in MCP (search_docs).
- Providing full file retrieval as a resource in MCP (read_doc).
- Developed with TypeScript.
Technology Stack
- TypeScript: Programming language for server implementation.
- @modelcontextprotocol/sdk: Used for MCP protocol implementation and communication.
- MiniSearch: Used for creating and querying full-text search indexes.
- wakachigaki: A lightweight tokenizer for Japanese text.
- fast-glob: Recursively explores Markdown files.
- Node.js fs/promises, path: For file reading and path handling.
Setup
- Install dependencies using npm.
- Create a docs directory and add Markdown files.
- Build with TypeScript.
Running the Server
Development
- You can run the server directly using TypeScript.
Production
- Run the built server.
The docs directory path can be omitted, defaulting to 'docs/'.
Tool: search_docs
- Description: Provides full-text search of Markdown files with high-precision Japanese tokenization.
Resource: read_doc
- Description: Retrieves full file content.
Logging
The server outputs detailed logs to stderr during major processes.
License
MIT