Skip to content

empeje/bakul-mcp

Repository files navigation

Bakul MCP Server

A Model Context Protocol (MCP) server that wraps the Bakul AI API endpoints.

Features

This MCP server provides tools for:

  • Health Check: Verify API availability
  • User Registration: Create new user accounts
  • API Key Management: Rotate authentication keys
  • Dataset Management: Create and update datasets
  • Public Dataset Access: Retrieve public datasets and schemas

Available Tools

Authentication

  • health_check - Check API health status
  • register_user - Register a new user account
  • rotate_api_key - Rotate API authentication key
  • get_api_key_status - Check API key configuration and test authentication

Dataset Management

  • create_dataset - Create a new dataset
  • update_dataset - Update an existing dataset
  • get_public_dataset - Retrieve a public dataset by username and ID
  • get_public_dataset_schema - Get the schema of a public dataset

Installation

pnpm install

Usage

Development

pnpm dev

Production

pnpm start

The server runs on stdio and communicates via the MCP protocol.

Configuration

The server connects to the Bakul API at https://ba.kul.to/api.

API Key Setup

  1. Get an API Key: Register a new user using the register_user tool
  2. Set Environment Variable:
    export BAKUL_API_KEY=your_api_key_here
  3. Alternative: Use the API_KEY environment variable

See config.example for a configuration template.

Environment Variables

  • BAKUL_API_KEY - Your Bakul API key (primary)
  • API_KEY - Alternative API key variable (fallback)

Tool Usage Examples

Each tool includes built-in usage instructions in its responses to help guide AI agents on proper usage patterns.

Technical Notes

  • Uses TypeScript with tsx for runtime execution
  • Zod schemas for input validation
  • Includes comprehensive error handling
  • All API responses include tool usage guidance for AI agents

API Reference

The server wraps endpoints documented in the Bakul API OpenAPI specification.

About

MCP for the missing storage layer for Personal AI: Bakul.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published