swarms

Independent review loop run before delivery in code work. An independent reviewer (Codex, or fresh Codex-style subagents) reads the whole PR against the approved outcome; the lead fixes in-scope functional findings and re-reviews until none remain. Its differentiator is independence and exhaustiveness — a reviewer that fails differently from the authors, run to clean — distinct from the recursive-refinement ladder. Invoked by the team lead from the unified pre-ship gate (code-mode and /swarm:refine) after the ship steps.

Returns the universal governance spec for swarm team runs — hard rules, briefing templates, gate presentation contract, launch mechanics, and pulse setup. The canonical source: invoked by launch.md at Step 1 and by user-authored shortcut commands that cannot read launch.md directly. Per-gate constants live in swarm:gate-presentation.

Returns the frozen gate constants — question, header, option labels and descriptions, digest field-list, and preview content — for every catalog gate. Invoked by the team lead fresh on arrival at each gate and on pulse re-emissions, in the same turn as the render. The transport contract, three renders, partition rule, and authoring rubric live in swarm:workflow-rules.

Code mode operational spec for the team lead. Returns lead identity, facilitator identity, mode-specific rules, suggest-members guidance, and phase arc for code-mode teams.

General mode operational spec for the team lead. Returns lead identity, facilitator identity, suggest-members guidance, and phase arc for general-purpose teams. The silent fallback when outcomes fit no specific mode, and a wrapper base for custom workflows — it has no shortcut command.

Triage mode operational spec for the team lead. Returns lead identity, facilitator identity, mode-specific rules, suggest-members guidance, and phase arc for triage teams that diagnose an issue without changing it.

Writing mode operational spec for the team lead. Returns lead identity, facilitator identity, ownership boundaries, editorial baseline, suggest-members guidance, and phase arc for writing-mode teams.

Evaluates ambient context artifacts (CLAUDE.md, memory, local skills, settings hooks) for compatibility with swarm governance. Returns a classified report so users can address interference before launching a team.

Recommends team composition based on outcomes and mode. Invoked by launch commands during team setup.

Detects when a stated outcome names a specific instance — a mechanism-as-means or a single named product — where a category was meant, and returns a neutral fork for the lead to show the user, or nothing. Invoked by launch commands at outcome capture, in place of the verbatim echo.

Resolves stuck review findings via a put-up-or-concede exchange. Invoked by the facilitator when a finding survives arbitration without new evidence.

Builds measurable validation criteria when no existing standards apply. Invoked by team agents during review.

Structural pattern analysis for writing-mode review. Performs paragraph-level decomposition and pattern aggregation. Invoked by the team lead during editorial review.