swarm

Swarm

Fan out N parallel background workers. They may cover separate slices, race the same brief, or mix both. The parent waits, aggregates, and returns one report.

Start

Open a todolist with one entry per phase before launching anything.

1. Frame
2. Fan out
3. Aggregate
4. Report

Phase A: Frame

1. State the done predicate and the artifact or report the swarm must return.
2. Choose the shape. Partition into slices, race N workers on identical briefs, or mix both. For a race or mixed shape, declare first pass, rank all, or best-of before spawning.
3. Set N from the user or derive it from the shape. N is total workers, not a concurrency limit.
4. The worker model is sonnet (the swarm workers row of the model table in the poteto-mode skill). For a model race, name each arm's model up front from sonnet, opus, haiku, and fable.
5. Give each worker its own writable output when it writes. When workers verify or measure commits, each brief names the exact SHAs. A measurement brief also names the method (sample count, what one sample is, order). The worker records both in its result.

Phase B: Fan out

Spawn all N workers in one message with subagent_type: "general-purpose", run_in_background: true, isolation: "worktree" when a worker writes, and the step 4 model.

When a worker must start from a non-default pushed branch, tell it the branch in the brief and have it check that branch out in its worktree first.

Every brief stands alone. Include the goal, scope, exact slice or race arm, how to verify, and what to report. Reports use PASS, ISSUES, or BLOCKED with evidence. A worker that can prove a defect reports ISSUES and lists every issue it can prove, not only the first.

If a worker drops out, proceed with N-1 and note it.

For fan-out that also plans the waves from a scope list, the claudefiles spawn skill owns that loop. The orchestrator skill holds a standing cross-branch coordinator. This skill stays the one-shot fan-out with a single aggregated report.