Facilitates communication and coordination between multiple AI agents through shared channels and persistent messaging.
Sponsored
mcp-broker addresses the limitations of existing multi-agent systems by enabling sophisticated inter-agent communication. It allows AI agents, such as multiple Claude instances, to discover each other, send asynchronous messages, engage in group discussions via channels, and coordinate complex tasks across sessions, with all communication data persistently stored in an SQLite database. This robust framework empowers developers to build more collaborative, long-running, and adaptable multi-agent applications than previously possible with basic agent interaction models.
주요 기능
01Role-based and Broadcast Messaging for targeted communication
02Group Communication Channels with history (create_channel, join_channel)
03Agent Profile Management and Spawner for automated agent instantiation
04Asynchronous and Persistent Messaging (send_message, poll_messages, get_history)
050 GitHub stars
06Agent Discovery and Status Tracking (list_peers, heartbeat)
사용 사례
01Distribute and coordinate complex code review tasks among specialized agents in parallel.
02Manage long-running projects by allowing tasks to be handed off and continued across different agent sessions.
03Implement supervisor-worker architectures where a central agent monitors and manages multiple specialized worker agents.
