Swarm

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. OMP runs at most task.maxConcurrency of them at once (32 by default) and queues the rest. A race that needs every arm to start together keeps N within that limit. Workers are leaves. They never call task, which keeps the swarm inside task.maxRecursionDepth (2).
4. Pick the worker agent. The default is task. Use scout for read-only coverage and exploration slices, and sonic for fully specified mechanical slices. Never name a model in a task item. Each agent's model comes from task.agentModelOverrides and modelRoles in ~/.omp/agent/config.yml, which /skill:setup-pstack (skill://setup-pstack) writes. For a model race, make each arm a distinct agent with a distinct operator override, and name each arm's agent up front. Claim that arms ran on different models only from the resolved-model metadata OMP returns with each result. Set each item's effort for the job only when the task schema shows an effort field (task.enableEffort). Otherwise omit it.
5. Give each worker its own writable output when it writes. Use a separate git worktree or directory per worker, and set isolated on the item only when the task schema shows it (task.isolation.enabled). Never depend on isolated being present. 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.