AnyGPT Router

⚠️ WORK IN PROGRESS: This package is under active development. APIs and routing mechanisms may change significantly. Use at your own risk in production environments.

A secure, enterprise-ready router for AI providers with flexible configuration and MCP protocol support.

🎯 Overview

The AnyGPT Router provides a unified interface to AI providers with:

• 🔧 Flexible configuration - Provider-based configuration system
• 🔒 Enterprise security - Secure credential management and proxy support
• 🎭 Provider abstraction - Consistent API across different AI providers
• 📊 MCP compliance - Seamless integration with MCP clients
• 🧪 Test-driven - Comprehensive test coverage
• ⚡ TypeScript-first - Full type safety and IntelliSense

🚀 Quick Start

Installation

npm install @anygpt/router

Basic Usage

import { GenAIRouter } from '@anygpt/router';
import { OpenAIConnectorFactory } from '@anygpt/openai';

// Create router and register connector factory
const router = new GenAIRouter();
router.registerConnector(new OpenAIConnectorFactory());

// Create connector instance
const connector = router.createConnector('openai', {
  apiKey: process.env.OPENAI_API_KEY,
  baseURL: 'https://api.openai.com/v1',
});

// Make request
const response = await connector.chatCompletion({
  messages: [{ role: 'user', content: 'Hello!' }],
  model: 'gpt-4o',
});

Configuration-Driven Approach

import { defineConfig, createRouter } from '@anygpt/router';

// Define configuration
const config = defineConfig({
  providers: {
    openai: {
      type: 'openai',
      apiKey: process.env.OPENAI_API_KEY,
      baseURL: 'https://api.company.com/openai/v1',
    },
  },
});

// Create router with config
const router = createRouter(config);

Documentation

| Document | Description | | ---------------------------------------------- | --------------------------------------------------------------- | | Configuration Guide | Provider configuration system, environment setup, customization | | API Reference | Complete API documentation, types, and examples | | Architecture | System design, patterns, and extension points | | Connector Usage | Provider-specific usage and best practices | | Logging Guide | Logger integration, custom loggers, and best practices |

Supported Providers

| Provider | Models | Features | | ------------- | -------------------------- | ----------------------------------------- | | OpenAI | GPT-4o, GPT-4, GPT-3.5, o1 | Chat completion, function calling, vision | | Anthropic | Claude Sonnet, Opus, Haiku | Native API, function calling | | Mock | test-model | Development and testing |

Configuration Examples

Environment-Based

export OPENAI_API_KEY=sk-your-key
export ANYGPT_LOG_LEVEL=info

Company Proxy

import { defineConfig } from '@anygpt/router';

const config = defineConfig({
  providers: {
    openai: {
      type: 'openai',
      apiKey: process.env.OPENAI_API_KEY,
      baseURL: 'https://api.company.com/openai/v1',
      headers: { 'X-Company-ID': 'engineering' },
    },
  },
});

Azure OpenAI

import { defineConfig } from '@anygpt/router';

const config = defineConfig({
  providers: {
    azure: {
      type: 'openai',
      apiKey: process.env.AZURE_OPENAI_API_KEY,
      baseURL: 'https://resource.openai.azure.com/openai/deployments/gpt-4o',
      headers: { 'api-version': '2024-02-15-preview' },
    },
  },
});

Architecture

MCP Client → @anygpt/mcp → @anygpt/router → AI Provider APIs

• Router Layer (this package): Provider abstraction, routing, security
• MCP Layer (separate package): MCP protocol implementation
• Connector Pattern: Extensible provider implementations
• Configuration System: Flexible provider-based configuration

Development

Setup

npm install
npm test          # Run tests
npm run build     # Build package

Testing

• Comprehensive test coverage for all functionality
• Mock connector for development and testing
• Type-safe test utilities and fixtures

Project Structure

src/
├── types/           # Core type definitions
├── connectors/      # Provider implementations
│   ├── base/        # Base connector interface
│   ├── openai/      # OpenAI connector implementation
│   └── registry.ts  # Connector registry
├── lib/             # Core router implementation
└── config.ts        # Configuration system

Key Features

Flexible Configuration

• Provider-based configuration system
• Easy customization and extension
• Environment variable support
• Company-specific configurations

Enterprise-Ready

• Secure credential management
• Proxy and custom header support
• Comprehensive error handling
• Production-tested patterns

Developer Experience

• Full TypeScript support
• Comprehensive documentation
• Test-driven development
• Clear error messages

🔮 Roadmap

• Additional Providers: Anthropic, Google, local models, custom endpoints
• Advanced Features: Request middleware, multi-provider fallback
• Monitoring: Metrics, logging, cost tracking
• Enhanced MCP Integration: Advanced MCP server features

License

MIT - See LICENSE for details.

Built with ❤️ for the MCP ecosystem | Documentation | API Reference | Architecture