Vibe Check MCP is a server implementing Anthropic's Model Context Protocol (MCP) that helps keep AI agents on track by providing meta-cognitive feedback. It acts as a "mentor layer" that interrupts agents when they're about to go down problematic paths, preventing over-engineering and keeping them aligned with user intentions.
You can run the server directly using npx without a local installation. This requires Node.js version 20 or higher.
npx -y @pv-bhat/vibe-check-mcp start --stdio
When successful, you'll see [MCP] stdio transport connected indicating the process is waiting for client connections.
Add this configuration to your MCP-aware client (Claude Desktop, Cursor, etc.):
{
"mcpServers": {
"vibe-check-mcp": {
"command": "npx",
"args": ["-y", "@pv-bhat/vibe-check-mcp", "start", "--stdio"]
}
}
}
npx -y @pv-bhat/vibe-check-mcp start --http --port 2091
You can verify the service is running with:
curl http://127.0.0.1:2091/health
For API interactions, send JSON-RPC requests to http://127.0.0.1:2091/rpc.
Vibe Check MCP supports several MCP-aware clients:
The server can be installed to Claude Desktop's configuration:
npx -y @pv-bhat/vibe-check-mcp install --client claude
Remember to restart Claude Desktop after installation for changes to take effect.
For the Cursor editor:
npx -y @pv-bhat/vibe-check-mcp install --client cursor
If your config is in a non-standard location, specify it:
npx -y @pv-bhat/vibe-check-mcp install --client cursor --config /path/to/mcp.json
npx -y @pv-bhat/vibe-check-mcp install --client windsurf --http
The --http flag is recommended for Windsurf as it uses HTTP transport by default.
npx -y @pv-bhat/vibe-check-mcp install --client vscode --config .vscode/mcp.json
If you don't specify a config path, the command will display a JSON snippet and a vscode:mcp/install link you can use directly.
Create a .env file with API keys for the LLM providers you plan to use:
# Gemini (default)
GEMINI_API_KEY=your_gemini_api_key
# Optional providers
OPENAI_API_KEY=your_openai_api_key
OPENROUTER_API_KEY=your_openrouter_api_key
ANTHROPIC_API_KEY=your_anthropic_api_key
# Optional overrides
DEFAULT_LLM_PROVIDER=gemini # accepts: gemini | openai | openrouter | anthropic
DEFAULT_MODEL=gemini-2.5-pro
Vibe Check MCP provides these key tools that agents can call:
The primary tool for reflection and oversight. Call it before taking significant actions:
// Example agent call
await tools.vibe_check({
task: "Optimize the database schema",
plan: "I'll modify the schema directly in production",
taskContext: "User asked for performance improvements"
});
Optional tool to log mistakes and fixes for future reference:
// Example agent call
await tools.vibe_learn({
sessionId: "user-session-123",
mistake: "Attempted to modify production database directly",
resolution: "Created migration script and staging test first"
});
Set rules for a specific session that the CPI layer will enforce:
// Set/update rules
await tools.update_constitution({
sessionId: "user-session-123",
rules: ["never write secrets to disk", "prefer unit tests before refactors"]
});
// Check current rules
const constitution = await tools.check_constitution({
sessionId: "user-session-123"
});
// Reset rules
await tools.reset_constitution({
sessionId: "user-session-123"
});
To get the most from Vibe Check, include instructions in your agent's system prompt:
As an autonomous agent you will:
1. Call vibe_check after planning and before major actions.
2. Provide the full user request and your current plan.
3. Optionally, record resolved issues with vibe_learn.
The research shows optimal "interrupt dosage" is about 10-20% of the agent's processing steps. Focus on calling vibe_check at key decision points rather than for every minor action.
To remove Vibe Check MCP from a client configuration:
vibe-check-mcp entryEach client configuration will have entries tagged with "managedBy": "vibe-check-mcp-cli" for easy identification.
AI Agent
To add this MCP server to Claude Code, run this command in your terminal:
claude mcp add-json "vibe-check" '{"command":"node","args":["/path/to/vibe-check-mcp/build/index.js"],"env":{"GEMINI_API_KEY":"YOUR_GEMINI_API_KEY"}}'See the official Claude Code MCP documentation for more details.
There are two ways to add an MCP server to Cursor. The most common way is to add the server globally in the ~/.cursor/mcp.json file so that it is available in all of your projects.
If you only need the server in a single project, you can add it to the project instead by creating or adding it to the .cursor/mcp.json file.
To add a global MCP server go to Cursor Settings > Tools & Integrations and click "New MCP Server".
When you click that button the ~/.cursor/mcp.json file will be opened and you can add your server like this:
{
"mcpServers": {
"vibe-check": {
"command": "node",
"args": [
"/path/to/vibe-check-mcp/build/index.js"
],
"env": {
"GEMINI_API_KEY": "YOUR_GEMINI_API_KEY"
}
}
}
}To add an MCP server to a project you can create a new .cursor/mcp.json file or add it to the existing one. This will look exactly the same as the global MCP server example above.
Once the server is installed, you might need to head back to Settings > MCP and click the refresh button.
The Cursor agent will then be able to see the available tools the added MCP server has available and will call them when it needs to.
You can also explicitly ask the agent to use the tool by mentioning the tool name and describing what the function does.
To add this MCP server to Claude Desktop:
1. Find your configuration file:
~/Library/Application Support/Claude/claude_desktop_config.json%APPDATA%\Claude\claude_desktop_config.json~/.config/Claude/claude_desktop_config.json2. Add this to your configuration file:
{
"mcpServers": {
"vibe-check": {
"command": "node",
"args": [
"/path/to/vibe-check-mcp/build/index.js"
],
"env": {
"GEMINI_API_KEY": "YOUR_GEMINI_API_KEY"
}
}
}
}3. Restart Claude Desktop for the changes to take effect