war-room

Multi-agent coordination prevents duplicate edits by applying per-task file locks. This ensures exclusive work on a given scope, stopping concurrent terminals from overwriting each other's changes.

Atomic task claiming is a multi-agent coordination mechanism that locks a specific task scope. It ensures only one agent works on a task at a time, preventing duplicated work across shared environments.

Yes, multi-agent coordination works across Claude Code terminals on different machines. It uses a shared war room to provide shared context, event logging, and agent discovery for team visibility.

To start coordinating tasks, tell your Claude Code terminal to connect to the war room. You should then run check_mission_board() before starting any task to claim atomic tasks and review shared context.

No specific dependencies are required to enable multi-agent coordination and file locks. The system enforces a defined frontmatter with name and description to guide activation alongside optional resource directories.

Use a war room when multiple agents need real-time collaboration with status broadcasts and auditability. It prevents duplicated work by managing atomic task claims and shared context, which manual assignment lacks.