Agent Forge
Build powerful AI agents
Agent Forge is a modern TypeScript framework for building AI agent systems using a powerful decorator-based architecture. Define agents with simple decorators, integrate seamlessly with LLM providers, add capabilities with tool decorators, and orchestrate complex multi-agent workflows. Experience the future of declarative AI development!

Decorator-Driven Architecture
Build AI agents using modern TypeScript decorators. Simply annotate your classes with @agent, @tool, and @llmProvider decorators for powerful, declarative agent development with zero boilerplate.
Seamless LLM Integration
Connect with any LLM provider using the @llmProvider decorator. Supports OpenAI, Anthropic, Azure, and custom providers with automatic dependency injection and configuration management.
Rich Tool Ecosystem
Add capabilities to agents with @tool decorators and MCP integration. Access web search, file systems, databases, and custom APIs. Build tool-enabled agents with just a decorator!
Advanced Team Orchestration
Create sophisticated multi-agent systems with teams and workflows. Agents collaborate intelligently under manager coordination to solve complex, multi-step problems efficiently.
Distributed Agent Networks
Build distributed agent systems with @a2aClient and @a2aServer decorators. Enable agents to communicate across networks with built-in authentication, load balancing, and fault tolerance.
Knowledge Base Integration
Add retrieval-augmented generation with @RAGChromaDb decorators. Agents automatically access document collections, embeddings, and knowledge bases for informed, context-aware responses.
Built-in Monitoring & Analytics
Monitor agent performance with @Visualizer and @RateLimiter decorators. Get real-time insights, cost tracking, performance metrics, and interactive timeline visualizations out of the box.
Plugin Architecture
Extend the framework with @plugin decorators for logging, security, caching, and custom functionality. Create reusable cross-cutting concerns that work across all agents automatically.