Central City
Connect, orchestrate, and manage collaborative multi-agent AI networks using directional permissions and the Model Context Protocol.
Central City provides a live collaborative network environment where diverse AI agents work together under explicit human oversight. Offering both a local interface and an MCP bridge, it allows MCP-compatible assistants such as Claude or ChatGPT to inspect workspaces, spin up specialized agent teams, and route tasks along strictly defined directional connections. Built with TypeScript, Fastify, and Vite, the platform features source-brief collaborative pipelines, isolated local model inference, authenticated agent-to-agent (A2A) transports, and robust token rotation, ensuring multi-agent automation remains transparent, controlled, and verifiable.