vibe-check-mcp-server

MCP.Pizza Chef: PV-Bhat

A second opinion for long-running AI work. Before your assistant charges further down a path, it can pause, describe the plan, and get blunt mentor-style feedback about whether things are getting too complicated. The authors have ended new feature work and publish only security and bug fixes now, with version 2.9 as the current release. You supply a key from Google, OpenAI, Anthropic, or OpenRouter, since the feedback itself is written by a model.

Coding

Other

Use This MCP server To

Catch my assistant before it over-engineers a simple fix Get a blunt second opinion on a shaky plan Interrupt a long agent run that lost the thread Sanity-check a vague request before any work starts Keep a coding agent on the smallest workable path

README

Vibe Check MCP

This project is in maintenance mode. Active feature development has ended; only maintenance patches (security and bug fixes) are published. v2.9.0 is the latest maintenance release. The server remains fully functional. Community forks and contributions are welcome under the MIT license.

KISS overzealous agents goodbye. Plug & play agent oversight tool.

Based on research:
In our study agents calling Vibe Check improved success +27% and halved harmful actions -41%

MIT License

_{Featured on PulseMCP “Most Popular (This Week)” • 5k+ monthly calls on Smithery.ai • research-backed oversight • STDIO + streamable HTTP transport}

vibe-check-mcp-server FAQ

Is this still maintained?

It is in maintenance mode. Only security and bug fixes ship, and version 2.9 remains current and fully working as of August 2026.

Do I need a paid key?

Yes. A key from Google Gemini, OpenAI, Anthropic, or OpenRouter, because the feedback comes from a model.

Which apps does it work in?

Claude Desktop, Cursor, Windsurf, VS Code, and other tools that support MCP servers.

Can I use this to review a plan before any code gets written?

Yes, that is its main purpose, and the authors recommend calling it early rather than after things go wrong.

How hard is setup?

One npx command plus a short block in your app's settings, and your chosen key in a file.

Is there evidence it helps?

The authors published a study reporting a 27% higher success rate and 41% fewer harmful actions in their own testing.