Fleet 1.13:Teams are now shipping 5x more PRs with autonomous pipelines.See what's new →
FleetFleet
CLI reference

fleet agent

fleet agent <list|get|create|start|stop|rollback|budget|logs|...> [<name>] [flags]

The primary command for managing individual agents: creating them from the catalog, listing, starting and stopping, inspecting budget utilization, viewing logs, and rolling back configuration to a previous revision. Agents run as Claude Code sessions inside tmux, and this command drives their lifecycle.

A new agent must be created first (fleet agent create or fleet template spawn) before fleet agent start <name> can launch it. Starting an agent is license-gated; read-only operations like list and budget are not.

Flags

--allWith start or stop, apply to every agent in the project (start --all launches all agents)
--enabledWith start --all, limit to agents whose config has enabled: true
--team <name>With start or stop, restrict the operation to agents in the named team
--all-projectsWith list, list every agent in the database across all projects instead of just the cwd project
--revision <n>With rollback, restore the agent config to a specific revision number (default: the previous revision)
--jsonEmit list or budget data as JSON

Examples

Start all enabled agents in the current repo

fleet agent start --all --enabled

Stop a single agent by name

fleet agent stop frontend-dev

Show per-agent budget utilization (cumulative run duration, not tokens)

fleet agent budget
# NAME          RUNS  DURATION  BUDGET    USED
# frontend-dev  12    1h14m     5h00m     25%

Roll an agent's config back to a specific revision

fleet agent rollback tech-lead --revision 3

Frequently asked questions

How do I create an agent before starting it?

Run `fleet agent create --name <n> --vendor claude-code --template <catalog-name>`, or `fleet template spawn <id> --name <n> --vendor claude-code`. `fleet agent start <name>` only launches an agent that already exists in the database.

What does the budget command measure?

Agent budgets are cumulative run duration in seconds, enforced against total run time. Fleet does not meter or cap tokens.

Run your first agent fleet

One binary. Five minutes. See every agent, coordinate every handoff, and keep a full audit trail of what your fleet did.