Neura Market
mrexodia/user-feedback-mcp logo

mrexodia/user-feedback-mcp

Free

简单的 MCP 服务器，在 Cline 和 Cursor 等工具中启用人机回圈工作流。

FreeFree tier
Type
Open Source
Visit Website GitHub (40 stars)

About mrexodia/user-feedback-mcp

User Feedback MCP is a simple Model Context Protocol (MCP) server that enables a human-in-the-loop workflow in AI coding tools like Cline and Cursor. It allows the AI to ask for user feedback before marking a task as completed, which is especially useful for developing desktop applications that require complex user interactions to test. The server provides a user_feedback tool that accepts a project directory and a summary of changes, prompting the user to approve or reject the action. It supports configuration via a .user-feedback.json file, which can specify a command to run and whether to execute it automatically. Installation is straightforward using uv, and a development mode is available for testing via uv run fastmcp dev server.py.

Key Features

Provides a user_feedback MCP tool for requesting user approval before task completion
Supports configuration via .user-feedback.json file with command and auto-execute options
Integrates with prompt engineering to ensure AI uses feedback tool automatically
Easy installation using uv (pip or curl script) from GitHub repository
Includes a development mode with web interface for testing MCP tools
Works with Cline and Cursor as well as any MCP-compatible tool
Open source with MIT license

Pros & Cons

Pros
  • Enables human oversight in automated AI workflows, increasing safety and reliability
  • Simple to set up with a single uv command and minimal configuration
  • Configurable auto-execution for commands after user approval
  • Works with multiple popular AI coding tools that support MCP
  • Open source and lightweight (Python-based MCP server)
Cons
  • Requires local installation and configuration of uv and the server
  • Limited to tools that support the Model Context Protocol (primarily Cline and Cursor)
  • Manual prompt engineering may be needed to ensure the AI consistently requests feedback
  • Not suitable for fully autonomous workflows as it requires human intervention

Best For

Developing desktop applications that require complex user interactions to testEnsuring AI agents obtain human approval before executing critical changesCreating safe AI-assisted workflows where tasks are reviewed by a human before completionIntegrating user feedback loops into AI coding assistants like Cline and Cursor

FAQ

What is User Feedback MCP?
It is a simple MCP server that enables a human-in-the-loop workflow in tools like Cline and Cursor, allowing the AI to ask for user feedback before completing a task.
How do I install User Feedback MCP?
First install uv globally (pip install uv or curl script), then clone the repository, and add the MCP server configuration to your cline_mcp_settings.json with the command 'uv --directory <path> run server.py'.
What is the .user-feedback.json file for?
It is a configuration file that can specify a command to run and whether to execute it automatically. If execute_automatically is enabled, the command will run instantly without needing to click Run manually.
How do I use the user_feedback tool?
The tool takes two arguments: project_directory and summary. It prompts the user for feedback on the changes described in the summary before the AI marks the task as completed.
What tools are compatible with this MCP server?
It is designed for Cline and Cursor, but any tool that supports the Model Context Protocol (MCP) should be able to use it.