Copilot Leecher

You can extend your Copilot workflow with a free MCP tool that waits for your feedback after a task is completed. This server lets you review and refine Copilot outputs within the same premium request, reducing the number of premium calls you need to make and enabling iterative improvements through a local web dashboard.

How to use

Use the MCP server with your Copilot client to submit a single, comprehensive task and then refine the result through an integrated review loop. When your Copilot agent finishes work, it calls a review tool exposed by the MCP server. Your feedback, entered in a local web UI, is returned as a tool call result and fed back into the same request to continue processing until you approve the work.

How to install

Prerequisites: you need Node.js and npm installed on your machine. You should also have a Copilot-enabled development environment ready to run an MCP client.

git clone https://github.com/user/copilot-leecher.git
cd copilot-leecher
npm install
npm run build

Configuration and usage options

The MCP server exposes a web dashboard for reviewing and providing feedback. You can configure your Copilot client to connect to the MCP server as shown below.

# VS Code settings example
"github.copilot.chat.mcp.servers": {
  "copilot-leecher": {
    "command": "node",
    "args": ["/absolute/path/to/copilot-leecher/dist/index.js"]
  }
}

Starting the dashboard

After installing and configuring, restart your IDE. The MCP dashboard starts automatically and is available at the local address.

http://127.0.0.1:3456

Usage pattern in practice

1. You submit a single, comprehensive prompt to Copilot. 2) Copilot begins work and then calls request_review to pause for feedback. 3) Open the local dashboard at http://127.0.0.1:3456 and provide your feedback. 4) The server returns the feedback as a tool call result, and the agent continues work within the same premium request. 5) Repeat as needed until you approve the work.

Troubleshooting

If the dashboard does not appear, ensure you used an absolute path in your configuration and that the build output exists. Check the console or output panel in your development environment for MCP Server messages.

Security considerations

The review flow relies on a local dashboard and tool calls within the same premium request. Treat the dashboard as a trusted interface and avoid exposing it beyond your local machine. If you enable remote access, secure the endpoint appropriately.

Notes

This server supports two main connection methods: an HTTP dashboard accessible at a local URL and a local stdio configuration that launches the MCP server from your environment. Use whichever method best fits your setup and ensure both point to the same review workflow.

Available tools

request_review

Submit feedback with a taskId and summary, wait for expert feedback, and continue work within the same premium request until approved.