mcp-server

The MCP Server is designed to enhance the capabilities of AI assistants by allowing them to safely access and interact with web content. It provides tools for web searching and content processing, ensuring that AI assistants can retrieve current information and perform tasks effectively. The server can be run using Docker for simplicity and security, or locally for more direct control. It supports both network and direct connection modes, making it versatile for different use cases, such as integration with LibreChat or Claude Desktop.

Features

• Web Search: Allows AI assistants to search the web using SearXNG for up-to-date information and resources.
• Web Access: Enables AI assistants to access and process website content, converting pages to markdown or extracting links.
• Docker Support: Provides a Docker setup for easy and secure deployment.
• Network and Direct Connection: Supports both SSE for network connections and stdio for direct connections.
• Customizable User-Agent: Allows setting a custom User-Agent for web requests.

MCP Tools

• {'Search': 'Search the web via SearXNG for current information, specific resources, or to perform calculations.'}
• {'Web': 'Access websites and process their content. Can convert pages to markdown for easy reading, get the raw content, or extract links.'}

MCP Resources

• {'MCP Specification': 'https://spec.modelcontextprotocol.io/'}
• {'MCP Python SDK': 'https://github.com/modelcontextprotocol/python-sdk'}
• {'MCP Example Servers': 'https://github.com/modelcontextprotocol/servers'}

Usage with Different Platforms

Docker

yaml:docker-compose.yml
services:
  mcp-server:
    environment:
      - SEARXNG_QUERY_URL=http://searxng:8080
      - SSE_HOST=0.0.0.0
      - SSE_PORT=8080
      - USER_AGENT=MCP-Server/1.0 (github.com/tcpipuk/mcp-server)
    image: ghcr.io/tcpipuk/mcp-server/server:latest
    ports:
      - "8080:8080"
    restart: unless-stopped
    stop_grace_period: 1s

LibreChat

yaml:librechat.yaml
mcpServers:
  mcp-server:
    iconPath: "/path/to/icon.png"
    label: "MCP Web/Search"
    type: sse
    url: http://mcp-server:8080/sse

Local

bash
curl -LsSf https://astral.sh/uv/install.sh | sh
uv venv
source .venv/bin/activate
uv sync
export SEARXNG_QUERY_URL="http://your-searxng-instance.local:8080"
export USER_AGENT="CustomAgent/1.0"
mcp-server --sse-host 0.0.0.0 --sse-port 3001