fleet shutdown [--json]Stops ALL fleet processes globally: the repo watcher, every org watcher, and all fleet-managed tmux sessions across every project. It also marks agents stopped in the database for the current project. It is the clean way to halt the entire fleet at once.
This is destructive and cannot be undone — agents must be restarted manually afterwards. It always kills tmux sessions; there is no graceful-versus-force distinction. The only flag is --json.
| --json | Emit a machine-readable JSON summary of what was stopped instead of text |
Shut down all fleet components
fleet shutdown
# Watcher stopped (PID 71204)
# Killed session: forge-frontend-dev
#
# Shutdown complete. 2 processes stopped.Emit a structured summary for a script
fleet shutdown --json
# {"stopped":2,"actions":["Watcher stopped (PID 71204)","Killed session: forge-frontend-dev"]}No. shutdown always kills tmux sessions immediately; there is no graceful-versus-force distinction. The only flag is --json.
No. shutdown stops the repo watcher, org watchers, and tmux sessions. Stop the brain separately with `fleet brain stop` if needed.
One binary. Five minutes. See every agent, coordinate every handoff, and keep a full audit trail of what your fleet did.