ワンクリックで
orchestration
Multi-phase orchestration for planners and executors.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Multi-phase orchestration for planners and executors.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
| name | orchestration |
| description | Multi-phase orchestration for planners and executors. |
| origin | pi-crew |
| triggers | ["orchestrate this","coordinate tasks","run this multi-phase","dispatch workers","coordinate team"] |
Use this skill when orchestrating multi-phase tasks across pi-crew teams and workers.
You are the orchestrator — the coordinator, not the executor.
You decompose, dispatch, verify, and iterate. You do NOT edit code directly. If you find yourself opening a file to fix a typo "real quick," stop — spawn a worker instead.
Adapted from oh-my-pi's orchestrate command pattern for pi-crew context.
Do not yield control while work remains unfinished. Run every phase to completion. The orchestrator owns the full lifecycle — from first dispatch to final green gate.
Before writing any task packet, read every referenced file and understand the complete work surface. Enumerate the entire surface before assigning work — do not assign work before you fully understand the scope.
Every set of edits with disjoint file scope MUST ship as one batch. If 5 tasks edit 5 different files and are independent of one another, dispatch all of them at once. Never serialize what can be parallelized.
Subagents have no shared context. Each worker only knows what you write in the task packet. Include all necessary context, file paths, constraints, and acceptance criteria in every task.
Run appropriate gates between phases: typecheck, tests, lint. Do not skip verification — a red phase must not advance to the next phase.
Commit after each green phase. Never commit a red tree. Only commit when all gates pass. If the phase fails, fix it first.
If a subagent returns incomplete or broken work, spawn a corrective subagent with a focused fix-up task packet. Do not fix a worker's mistakes yourself — respawn a new worker to fix them.
Maintain the original scope exactly. Do not expand scope because you "spot more work," and do not shrink it because "it's good enough for now." If scope needs to change, escalate to the requester.
git status and git diff to understand current tree state.team call.requirements-to-task-packet skill).workspaceMode: 'worktree' when parallel edits risk conflict.Before launching a new phase, verify:
If ANY answer is NO → Stop. Complete planning before dispatching.
These are the behaviours that kill orchestration quality — avoid them:
| Anti-pattern | Why it's wrong |
|---|---|
| Editing files yourself "because it's faster" | You are the orchestrator, not an editor. Speed comes from correct delegation, not shortcutting. |
| Yielding after phase 1 with "ready to continue?" | The requester gave you a goal, not a conversation. Drive to completion. |
| Dispatching one subagent at a time when five could run in parallel | Wasted time. Enumerate first, then batch-dispatch all independent tasks. |
| Skipping typecheck/tests between phases | A red phase propagates errors forward. Always verify before advancing. |
| Marking todos done without verifying | Unverified work is undone work. Run the gate, check the output, then mark done. |
Use the team tool with appropriate action for dispatching work:
action: 'run' with a named team for multi-role work (implementation, review, research).inbox/outbox) for cross-worker coordination when workers need to signal completion or report blockers.| Concept | When to use |
|---|---|
team: 'implementation' | Complex multi-phase implementation with parallel specialists |
team: 'fast-fix' | Small targeted fixes, single-phase |
team: 'review' | Code review and security review phases |
team: 'research' | Investigation before implementation planning |
team: 'parallel-research' | Multi-project/source audits |
workflow: 'implementation' | Adaptive fanout where planner decides subagent allocation |
AGENTS.md and project-level config before planning phases.npm run check (requires prior build), ./test.shnpm test, npm run typechecknpm test, npm run test:allFor orchestration skill itself:
cd pi-crew
npx tsc --noEmit
node --experimental-strip-types --test test/unit/team-recommendation.test.ts
npm test
For orchestrated work: run the gate commands appropriate to the target subproject after each phase, and again after final phase.
Iterative multi-round codebase audit with diminishing-returns detection. Run 5-20+ rounds, each focusing on one specific area. Built from 19 rounds of dogfooding pi-crew on itself.
Pi TUI crew widget data sources, display priority, and rendering performance.
Spawn 3 adversarial subagents (Skeptic, Pragmatist, Critic) to evaluate a decision, architecture choice, or plan. Anti-anchoring: each role receives ONLY the question, not conversation history. Aggregates votes into consensus recommendation with dissent tracking. Use when facing critical decisions, architecture choices, security tradeoffs, or plan reviews where single-perspective analysis is insufficient.
Background worker, heartbeat, stale-run, crash-recovery, and deadletter workflow. Use when debugging stuck/dead workers or changing async run reliability.
Child Pi worker spawning, lifecycle callbacks, and failure modes.
Use when constructing worker prompts, reading artifacts/logs, summarizing runs, compacting context, or handing work between agents.