Multi-Agent Coordination
Build modular AI systems that scale. Compose multiple specialized agents with safe state sharing. Deterministic scheduling prevents race conditions. Agents can use other agents as tools. Built-in patterns for delegation, collaboration, and orchestration.
@agent
class ResearchAgent:
async def research(self, query: str):
# Use search agent as a tool
results = await self.use_tool(
SearchAgent, query=query
)
return await self.synthesize(results)
@agent
class OrchestratorAgent:
async def analyze(self, topic: str):
# Coordinate multiple agents
research = await ResearchAgent().research(topic)
analysis = await AnalysisAgent().analyze(research)
return analysisSafe State Sharing
Agents can share state safely without race conditions. Built-in coordination primitives ensure deterministic execution. No manual locking or synchronization needed.
Agents as Tools
Compose agents like functions. One agent can use another agent as a tool. Build specialized agents and orchestrate them to solve complex problems.
Delegation Patterns
Built-in patterns for common multi-agent scenarios. Delegate work to specialized agents. Aggregate results from multiple agents. Fan-out/fan-in orchestration.
Modular & Testable
Each agent is independently testable. Mock dependencies for unit tests. Build complex systems from simple, well-tested components.