crewai-agenium

CrewAI integration for the Agenium agent:// protocol — build truly distributed AI crews across the agent network.

Install

pip install crewai-agenium

Quick Start

from crewai_agenium import AgeniumAgent, AgeniumTask, AgeniumCrew

# Create agents from remote agent:// endpoints
researcher = AgeniumAgent(agent_uri="agent://researcher.agenium")
writer = AgeniumAgent(agent_uri="agent://writer.agenium")

# Build a distributed crew
crew = AgeniumCrew.from_agent_uris([
    {"uri": "agent://researcher.agenium", "task": "Research AI trends"},
    {"uri": "agent://writer.agenium", "task": "Write a report"},
])
result = crew.kickoff()

Components

Class	Description
AgeniumTool	Use agent:// endpoints as CrewAI tools
AgeniumAgent	Wrap agent:// as a CrewAI Agent
AgeniumTask	Delegate tasks to remote agents
AgeniumCrew	Distributed crew across agent:// network
AgeniumRegistrar	Register agents with DNS (API key: dom_...)
AgeniumCrewCallback	Stream crew events to a monitoring agent

Auto-Discovery

# Build crew from all agents with "research" capability
crew = AgeniumCrew.discover_and_build(
    capability_filter="research",
    task_description="Research quantum computing advances",
)

Hybrid Crews

Mix local and remote agents — see examples/hybrid_crew.py.

License

MIT

Bug Reporting

This project includes optional bug reporting to the Agenium monitoring server.

Configuration

Set the following environment variables to enable bug reporting:

BUG_REPORT_URL=http://130.185.123.153:3100
BUG_REPORT_TOKEN=your_token_here

Bug reporting is disabled by default — it only activates when BUG_REPORT_TOKEN is set. Reports are sent asynchronously (fire and forget) and never block the main application.