basemind-comms

To coordinate multiple agents working on the same repository, you can use a broker to exchange room-scoped conversations and per-agent inbox messages. This ensures agents stay aligned on start and finish decisions while maintaining context and accountability.

Room-scoped messaging for collaborative coding works by automatically joining agents to repo-scoped rooms. Agents use a two-tier message payload with frontmatter summaries and lazy-loaded bodies to keep conversations focused and predictable.

The best way to check if another agent is modifying the same code is to read recent activity from the agent inbox and list room history. This surfaces ongoing tasks and blockers before you start making changes.

Start collaborative coding tasks by running inbox read and room list commands to surface recent activity. Then post a concise starting message to the relevant room to announce your work and establish a linked reply thread.

You can maintain context and accountability across multiple agents by using room history and reply threads. These features trace collaborative decisions and ensure predictable interactions within workspace-scoped rooms.

Agent coordination requires a two-tier message payload format combining frontmatter summaries with lazy-loaded bodies. This structure keeps rooms focused and ensures predictable, traceable collaboration across the workspace.