user-feedback-mcpSimple MCP Server to enable a human-in-the-loop workflow in tools like Cline and Cursor.

What are the main limitations of this tool?

The server requires manual configuration and installation, depends on uv and other tooling, and is only suitable for desktop applications needing complex user interactions. It also needs to be run locally, which may limit scalability for larger projects.

Asked by Xavier Costa · Aug 9, 2025

Which tools can use the user-feedback MCP server?

It is designed for use with tools that support MCP, notably Cline and Cursor. By adding the server to their MCP configuration, these tools can invoke the `user_feedback` tool to request human input before marking tasks complete.

Asked by Zara Ahmed · Jun 10, 2025

What does the .user-feedback.json configuration do?

When you save the MCP configuration in Cline, a .user-feedback.json file is created containing keys like `command` and `execute_automatically`. If `execute_automatically` is true, the specified command runs immediately on startup, eliminating the need to click Run manually; otherwise you must trigger it manually.

Asked by Olamide Fashola · Jun 7, 2025

How do I install and run the user-feedback-mcp server?

First install uv globally (pip install uv on Windows or curl … on Linux/Mac). Clone the repo, then configure Cline’s MCP servers by adding an entry in cline_mcp_settings.json pointing to the repository and setting the command to run `uv args: [ directory c:\MCP\user feedback mcp run server.py ]`. Finally run the server with `uv run fastmcp dev server.py` to launch the web interface at http://localhost:5173.

Asked by Grzegorz Lewandowski · May 8, 2025