Vibe Check MCP Server for AI Workflow Oversight

.jpeg)

Overview

The Vibe Check MCP Server is a metacognitive oversight tool designed to prevent cascading errors in AI workflows. It introduces strategic pattern interrupts to ensure AI agents stay aligned with user intentions, avoiding over-engineering and scope creep. Built with LearnLM 1.5 Pro (Gemini API), it fine-tunes workflows for pedagogy and metacognition, ensuring AI agents don't get stuck in tunnel vision.

Key Features

• Pattern Interrupts: Breaks tunnel vision with metacognitive questioning.
• Plan Simplification: Encourages simplification of complex workflows.
• Self-Improving Feedback Loops: Logs mistakes to improve future decision-making.

The Problem: Pattern Inertia

AI agents often exhibit pattern inertia, where they continue down a reasoning path even when it’s clearly wrong. This leads to:
- Tunnel Vision: Stuck in one approach, unable to see alternatives.
- Scope Creep: Simple tasks evolve into enterprise-scale solutions.
- Overengineering: Adding unnecessary layers of abstraction.
- Misalignment: Solving a different problem than the one asked.

Tools for Metacognitive Oversight

🛑 vibe_check

A pattern interrupt mechanism that breaks tunnel vision with metacognitive questioning.

vibe_check({
  "phase": "planning",           // planning, implementation, or review
  "userRequest": "...",          // FULL original user request 
  "plan": "...",                 // Current plan or thinking
  "confidence": 0.7              // Optional: 0-1 confidence level
})

⚓ vibe_distill

A meta-thinking anchor point that recalibrates complex workflows.

vibe_distill({
  "plan": "...",                 // Detailed plan to simplify
  "userRequest": "..."           // FULL original user request
})

🔄 vibe_learn

A self-improving feedback loop that builds pattern recognition over time.

vibe_learn({
  "mistake": "...",              // One-sentence description of mistake
  "category": "...",             // From standard categories
  "solution": "..."              // How it was corrected
})

Installation & Setup

Installing via Smithery

Automatically install for Claude Desktop via Smithery:

npx -y @smithery/cli install @PV-Bhat/vibe-check-mcp-server --client claude

Manual Installation via npm

# Clone the repo
git clone https://github.com/PV-Bhat/vibe-check-mcp-server.git
cd vibe-check-mcp-server

# Install dependencies
npm install

# Build the project
npm run build

# Start the server
npm run start

Integration with Claude

Add to your claude_desktop_config.json:

"vibe-check": {
  "command": "node",
  "args": [\
    "/path/to/vibe-check-mcp/build/index.js"\
  ],
  "env": {
    "GEMINI_API_KEY": "YOUR_GEMINI_API_KEY"
  }
}

Environment Configuration

Create a .env file in the project root:

GEMINI_API_KEY=your_gemini_api_key_here

Agent Prompting Guide

Include these instructions in your system prompt:

As an autonomous agent, you will:
1. Treat vibe_check as a critical pattern interrupt mechanism
2. ALWAYS include the complete user request with each call
3. Specify the current phase (planning/implementation/review)
4. Use vibe_distill as a recalibration anchor when complexity increases
5. Build the feedback loop with vibe_learn to record resolved issues

When to Use Each Tool

API Reference

See the Technical Reference for complete API documentation.

Architecture

Vibe Check implements a dual-layer metacognitive architecture based on recursive oversight principles. Key insights:
1. Pattern Inertia Resistance: LLM agents naturally demonstrate a momentum-like property in their reasoning paths.
2. Phase-Resonant Interrupts: Metacognitive questioning must align with the agent's current phase.
3. Authority Structure Integration: Agents must treat external metacognitive feedback as high-priority interrupts.
4. Anchor Compression Mechanisms: Complex reasoning flows must be distilled into minimal anchor chains.
5. Recursive Feedback Loops: All observed missteps must be stored and leveraged to improve interrupt efficacy.

Documentation

Contributing

We welcome contributions! Check out our Contributing Guidelines to get started.

License

MIT