Status: [UNTESTED] Endpoints found: http://localhost:3000/api/docs http://localhost:3000/api/proxy http://localhost:3000/api/proxy/health http://localhost:3000/api/proxy/providers http://localhost:3000/api/proxy/prompt Source: https://github.com/LiteObject/demo-ai-proxy-service # AI Proxy Service A vendor-agnostic AI proxy service built with NestJS and TypeScript. Features retry mechanisms, structured logging, configuration validation, and vendor abstraction for seamless AI provider integration. ## Features ### Core Capabilities - **Vendor-Agnostic Architecture** - Provider Abstraction Layer (Strategy + Factory + Adapter) supports multiple AI providers (AWS, Azure, Google, OpenAI) - **Enterprise-Grade Reliability** - Intelligent retry logic with exponential backoff and jitter - **Configuration Validation** - Comprehensive schema validation with class-validator decorators - **Structured Logging** - JSON-formatted logs with performance tracking and correlation IDs - **Enhanced Error Handling** - Hierarchical exception system with retryability classification - **Async System Prompt Loading** - File-based caching with mtime validation for optimal performance ### AI Integration Features - **AWS Bedrock Integration** - Support for Claude, Titan, and other foundation models - **Specialized Safety Analysis** - Dedicated incident report analysis with expert safety prompts - **Model Discovery** - Comprehensive endpoint for provider and model information with pricing - **Configurable Model Settings** - Per-operation model, token, and temperature configuration - **Multi-Provider Support** - Ready for Azure OpenAI, Google Vertex AI, and OpenAI integration ### Production-Ready Features - **Input Validation** - Comprehensive DTO validation with security constraints - **Rate Limiting** - Configurable throttling (100 requests/minute by default) - **Health Monitoring** - Detailed health checks with provider status - **Security Hardening** - CORS, Helmet, input sanitization, and XSS protection - **Performance Metrics** - Operation tracking with success rates and latency percentiles - **API Documentation** - Interactive Swagger/OpenAPI documentation ## Architecture Overview ```mermaid graph TB Client[Client Applications] --> Controller[ProxyControl