npx mcpbar@latest install paybyrd/ai-agent-toolkit-js -c claude

npx mcpbar@latest edit -c claude

{
  "mcpServers": {
    "Paybyrd AI Agent Toolkit": {
      "command": "npx",
      "args": [
        "-y",
        "@paybyrd/ai-agent-toolkit-js"
      ],
      "env": {}
    }
  }
}

Paybyrd AI Agent Toolkit

A toolkit for integrating Paybyrd's payment processing capabilities with AI models like OpenAI and Anthropic Claude.

Overview

This toolkit enables AI assistants to perform payment operations through the Paybyrd API:

• Create payment links for customers
• Process refunds
• Retrieve order information

The toolkit uses a modular architecture with support for different AI platforms.

Repository Structure

This monorepo is organized into these packages:

• core (@paybyrd/ai-agent-toolkit): Core functionality and shared tools
• claude (@paybyrd/ai-agent-claude): Anthropic Claude integration using Model Context Protocol
• openai (@paybyrd/ai-agent-openai): OpenAI integration with function calling

Quick Start

Prerequisites

• Node.js 18 or higher
• Paybyrd API key
• OpenAI API key (for OpenAI integration) or Anthropic API key (for Claude integration)

Installation

Each package can be installed separately depending on which AI platform you're using:

# For the core functionality
npm install @paybyrd/ai-agent-toolkit

# For Claude integration
npm install @paybyrd/ai-agent-claude

# For OpenAI integration
npm install @paybyrd/ai-agent-openai

Examples

See the /examples directory for complete implementation examples:

• /examples/openai: Example implementation with OpenAI
• /examples/modelcontextprotocol: Example implementation with Claude using the Model Context Protocol

Available Tools

The toolkit provides these payment operations:

• Create Payment Link: Generate payment links to send to customers
• Create Refund: Process refunds for transactions
• Retrieve Order: Get details about an existing order

Development

# Install dependencies
npm install

# Build all packages
npm run build

# Run tests
npm test

License

MIT