MCP Web Utilities Server

# Technical Overview ## Project This project is a Model Context Protocol (MCP) server that exposes web search and webpage fetch tools for MCP clients (e.g., LM Studio). The server uses `FastMCP` and runs with the `stdio` transport. ## Documentation Philosophy To keep the agent's context clean, this overview is intentionally kept short. It serves as a central hub. Detailed information is modularized into specialized markdown files. **Always prefer reading the specific linked file related to your task rather than parsing the whole documentation.** ## Modules & Workflows - **Architecture**: Components and data flow: [docs/technical/architecture.md](./technical/architecture.md) ## Project Structure ```text Project/ ├── docs/ # Technical documentation ├── src/ # MCP server source code │ ├── mcp_server.py # FastMCP server and tool registrations │ ├── main.py # Entry point (stdio transport) │ └── utils/ │ ├── search.py # DDGS (Dux Distributed Global Search) helpers │ └── fetch_webpage.py # Webpage fetch to markdown (crawl4ai) ├── start_mcp_server.cmd # Convenience script to launch the server ├── AGENTS.md # Agent operating guidelines └── pyproject.toml # Dependencies and project metadata ``` ## Technology Stack - **Language**: Python 3.13+ - **MCP Server**: `FastMCP` (`mcp` package) - **Search**: `ddgs` (DDGS | Dux Distributed Global Search, multi-source) - **Web crawling**: `crawl4ai` (async) - **Async runtime**: `asyncio`