@pendulum-baas/mcp

MCP (Model Context Protocol) server for natural language querying of deployed Pendulum BaaS backends. This tool allows you to interact with your Pendulum backend using natural language queries through compatible AI assistants like Claude Desktop.

Installation

npm install -g @pendulum-baas/mcp

Usage

Arguments

• --url (required): The URL of your app's Application Load Balancer.
• --jwt-secret (required): JWT Secret generated during deployment — can be found in AWS Secrets Manager.

Run MCP Server Manually

You can start the MCP server with your Pendulum backend credentials locally:

pendulum-mcp --url <backend-url> --token <auth-token>

Configuration with Claude Desktop

Add the following to your Claude Desktop MCP configuration:

{"mcpServers": {
  "pendulum": {
    "command": "pendulum-mcp",
    "args": [
      "--url", "<your-alb-url>",
      "--jwt-secret", "<your-jwt-secret>"
    ],
    "env": {
      "PATH": "/Users/username/.nvm/versions/node/v22.14.0/bin:/usr/local/bin:/usr/bin:/bin"
    }
  }
}}

Available Tools

The MCP server provides the following tools for interacting with your Pendulum backend:

check-health-status

Get real-time insights into the health and status of both backend services.

get-active-users

Returns the number of users currently connected to your application.

infer-schema

Given a collection, infers and generates JSON that represents the schema of the collection.

validate-schema

Given a collection and an expected schema, identifies any documents in the collection that do not conform to the expected schema. Returns _id values for those that do not match.

start-realtime-logs

Monitor and log all realtime events as they are broadcast out to active users for a specified amount of time. Log file is timestamped and written to user's local /tmp directory.

stop-realtime-logs

Stop a real-time monitoring session.

realtime-logs-status

Check the status of any active real-time monitoring sessions.

Features

• Natural Language Queries: Ask questions about your backend in plain English
• Real-time Monitoring: Get live insights into your backend state
• Service Health: Monitor the health and performance of your Pendulum services
• Schema Inference and Validation: Have LLMs generate schemas based on your existing data so you can add stricter data validation rules to your frontend code.

Requirements

• Node.js 18.0.0 or higher
• Valid Pendulum backend with authentication
• Compatible MCP client (like Claude Desktop)

License

MIT