easypi/mcp-proxy-fetch repository overview

services:
  fetch:
    image: easypi/mcp-proxy-fetch
    command: "--host=0.0.0.0 --port=5757 -- mcp-server-fetch --ignore-robots-txt --user-agent=Mozilla/5.0 --proxy-url=http://192.168.1.1:8081"
    ports:
      - "5757:5757"
    restart: unless-stopped

services:
  mcp-inspector:
    image: ghcr.io/modelcontextprotocol/inspector
    ports:
      - "6274:6274"
      - "6275:6275"
    volumes:
      - ./data:/home/node/.mcp-inspector
    environment:
      - DANGEROUSLY_OMIT_AUTH=true
    restart: unless-stopped

$ chatgpt -n -i --tools --mcp http://127.0.0.1:5757/mcp
Entering interactive mode. Using thread 'cmd_6d16'. Multiline mode is OFF.
Commands: 'clear' (clear screen), 'multiline' (toggle multiline input), 'exit' or Ctrl+C (quit).

[2026-08-19 07:03:52] [Q1] [0] list tools
Here are the tools I have access to:

## Tools Available

| Tool | Description |
|------|-------------|
| **fetch** | Fetches a URL from the internet and optionally extracts its contents as markdown or raw HTML content. You control what you get back by setting parameters like max length and character start index. |

### Fetch Tool Details

**Parameters:**
- `url` (required) - The URL to fetch
- `max_length` (optional, default 5000) - Maximum number of characters to return
- `start_index` (optional, default 0) - Output starting position for truncated content
- `raw` (optional, default false) - When enabled, returns actual HTML without simplification; otherwise returns markdown format

This tool enables me to help you retrieve web page content, articles, documentation, or any publicly accessible URL content when needed.

---

Would you like me to fetch something specific using this tool?