Hand big work to a swarm.

Mini-bosses drive big work through phase gates until it is actually done.

• Scope big work only
• Bosses per-boss child cap
• Gates phase by phase
• View Lilly Pod

1. Describe the goal — state what done looks like, not each command.
2. Bosses fan out — capped child agents each take a slice.
3. Self-correction runs — stalls and failures rerun without you.
4. Gates clear in order — steer at the gate, not every agent.
5. Get the report — the boss finishes, then reports back.

The pod view turns background work into a readable topology — a whole swarm at a glance.

lilly pod — live swarms

Autonomous swarms with live counts and alerts.

Describe the big goal

• Give Toad the outcome, not each command — state what done looks like.
• The mini-boss handles the plan and the breakdown.
• A genuinely big goal lands as a swarm, not a single agent.

Mini-bosses orchestrate the swarm (per-boss child cap)

• Mini-bosses split the goal and spawn child agents, each taking a slice.
• Every boss runs under a per-boss child cap — scoped, not infinite.
• The brain runs on almost any model; the Gemini and Antigravity CLIs stay agents-only.

Self-correction runs automatically

• When a child stalls or fails, the boss reruns and repairs it on its own.
• Progress continues without you stepping in for every hiccup.

Phase gates keep it on track

• Each phase must clear before the next begins.
• The pod view shows children, progress, and alerts live; badges flag swarms needing a decision.
• Steer at the gate, not at every agent. Repo work reaches GitHub via the gh CLI.

Get completion visibility (each swarm stays tied to its workspace)

• The boss keeps going until the plan is done, then reports back.
• Outputs land in the right repo context — each swarm stays tied to its workspace.
• Calls to a confined remote agent stay approval-gated.
• Want to watch a slice live? Spawn an Agent View pane.

When to reach for it

When the job is too big to watch and too long to drive by hand.