Lightweight multi-agent orchestration with decorator-based composition
AgentForge is a lightweight Python framework for orchestrating multiple AI agents. Define agents with simple decorators, connect them via an event bus, and let them collaborate on complex tasks — no YAML configs needed.
- ✅ Decorator-based agent definition
- ✅ Built-in event bus for agent communication
- ✅ Sequential, parallel, and pipeline execution modes
- ✅ Pluggable LLM backends (OpenAI, Anthropic, local)
- ✅ Built-in retry and error handling
- 🔜 Web dashboard for monitoring
- 🔜 Persistent agent memory
pip install agentforge-aifrom agentforge import Agent, Orchestrator
@Agent(name="researcher")
def research(query: str) -> str:
return f"Research results for: {query}"
@Agent(name="writer")
def write(context: str) -> str:
return f"Article based on: {context}"
orch = Orchestrator()
orch.pipeline([research, write])
result = orch.run("quantum computing trends")graph TD
A[User Task] --> B[Orchestrator]
B --> C[Event Bus]
C --> D[Agent 1]
C --> E[Agent 2]
C --> F[Agent N]
D --> G[Result Aggregator]
E --> G
F --> G
- Python 3.11+ with asyncio
- Pydantic for configuration
- httpx for LLM API calls
This project was inspired by deer-flow and agency-agents but takes a simpler, decorator-first approach with no external configuration files.
Built by Officethree Technologies | Made with ❤️ and AI