---
name: swarm
description: "Fan out N parallel workers over separate slices or racing one brief, drain them, and return one report. Use for /swarm, 'swarm this', or parallel coverage, races, gauntlets, and exploration. For N attempts at the same artifact where a base is picked and the losers grafted in, use `arena` instead."
---

# Swarm

On Codex, the todolist below is `update_plan` and spawning workers is
`spawn_agent`; see `../poteto-mode/references/codex-tools.md`.

Fan out N parallel cloud 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 the cloud concurrency limit.
4. Pick the worker model from `swarm workers` in `plugins/pstack-nikki/models.json` when present; absent -> omit `model`. For a model race, name each arm's model up front.
5. Give each worker its own writable output when it writes. A worker that writes in the repo gets its own git worktree on its own branch, never a shared checkout; otherwise `.nikki-agents/swarm/<slug>/worker-<n>/`.

## Phase B: Fan out

Spawn all N workers in one message, one agent per worker. Pick the agent type from the work: any of the seven in the `poteto-mode` agent table (on Claude Code, `developer-codex` and `reviewer-codex` per its Codex-first rule), or an `orchestrator` running a role when a worker's slice is a whole workstream. Run them in the background where the harness allows, using the configured model. Run remote when available; fall back to local only when the worker needs access to something on the user's computer.

When a worker must start from a non-default branch, tell it which branch to check out 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.

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

## Phase C: Aggregate

Read the terminal results. For coverage, every required slice needs a result. For a race, apply the selection rule declared up front. Use first pass, rank all, or best-of. Do not paste raw worker dumps.

Keep a compact result table, one-line evidenced issues, and explicit gaps or dropouts.

## Phase D: Report

Return one consolidated in-chat report with the table, issue one-liners, gaps or dropouts, and the race rule when used.

<!-- dotai:models:start -->
## Models

Stamped from `plugins/pstack-nikki/models.json` (edit there, rerun `generate-models.py`). Row absent -> omit `model`, child inherits. A spawner reads the entry for its own harness.

- `swarm workers`: On Claude Code: `sonnet`. On Codex: `gpt-5.6-terra`. On Copilot CLI: `claude-sonnet-5`, `gpt-5.5`.
<!-- dotai:models:end -->