소개
Agent Switchboard is designed to streamline fragmented agent communication, providing a robust solution for connecting various AI agents, human tools, and other systems. By integrating the Model Context Protocol (MCP) for standardized agent-to-tool communication with Redis Streams, it offers a fast, durable, and observable message bus. This setup ensures agents can collaborate effectively, manage tasks with durable logs, consumer groups, and acknowledgements, and leverage a lightweight, yet powerful, communication fabric for multi-agent orchestration.
주요 기능
- Production-ready Node.js MCP server exposing Redis as a durable agent bus.
- 0 GitHub stars
- Docker Compose for simplified one-command deployment of Redis, MCP, and the bridge.
- Standardized agent-tool interaction via Model Context Protocol (MCP) for major LLMs.
- REST bridge (FastAPI) for integrating non-MCP agents via HTTP.
- Durable agent communication and replay using Redis Streams with consumer groups.
사용 사례
- Facilitating multi-agent collaboration and orchestration in complex workflows.
- Providing a durable and observable message bus for agent tasks, results, heartbeats, and audit trails.
- Standardizing communication between LLM agents (Claude, Gemini, Copilot) and external tools or data sources.