Model Context Protocol (MCP) Website Fetcher

A Model Context Protocol (MCP) server implementation that provides a simple website fetching service. This server exposes a tool that allows clients to fetch and retrieve the content of any website through the MCP protocol.

What is MCP?

Model Context Protocol (MCP) is a protocol that enables LLMs to interact with external tools and resources. MCP servers can provide three main types of capabilities:

1. Resources: File-like data that can be read by clients (like API responses or file contents)
2. Tools: Functions that can be called by the LLM (with user approval)
3. Prompts: Pre-written templates that help users accomplish specific tasks

This implementation focuses on providing a tool capability for website fetching.

Local Development

1. Clone the repository:

   git clone <your-repository-url>
   cd <repository-name>

2. Install dependencies:

   pip install -r requirements.txt

3. Run the MCP server (Streamable HTTP):

   python app.py --port 8000 --path /mcp

The server will start a Streamable HTTP MCP endpoint at http://localhost:8000/mcp.

Notes

• This server uses the Streamable HTTP transport. Clients should connect with transport set to http.
• In production, validate Origin headers and use HTTPS.

Docker

Local Docker Build

1. Build the Docker image:

   docker build -t mcp-website-fetcher .

2. Run the container:

   docker run -p 8000:8000 mcp-website-fetcher

Using GitHub Container Registry

The application is automatically built and pushed to GitHub Container Registry (GHCR) on every push to the main branch. You can pull and run the latest image using:

docker pull ghcr.io/<your-github-username>/<repository-name>:main
docker run -p 8000:8000 ghcr.io/<your-github-username>/<repository-name>:main

Available container images

A ready-for-use container image can be obtained by docker pull otomato/mcp-website-fetcher