메뉴
Use to run a rigorous multi-lane review before declaring work done — for any broad, risky, shared, security-sensitive, or release-facing change. Uses native scheduling for independent review lanes and gates on ALL of them passing with a single binding verdict.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use BEFORE implementing anything non-trivial — when a request is broad, multi-step, ambiguous, or risky and needs a decision-complete plan. Runs explore-first research, an interview on genuine unknowns, writes and self-reviews one plan, then stops at an explicit approval gate.
Activate when the user types `csw` (alone, or as a prefix like `csw <task>`), or asks to "run the loop" / work with full rigor. Runs the evidence-bound autonomous loop — bind a goal, plan if needed, then drive each criterion through test-first implementation, real manual QA, independent review, and paired cleanup until the completion oracle passes.
Use to EXECUTE an approved plan with discipline. Resumes from durable .csw/ goal state, drives each task through test-first implementation, real manual QA, independent verification, and paired cleanup, and only finishes when the completion oracle passes. Bootstraps csw-plan first if no plan exists.
Use when work contains two or more INDEPENDENT subtasks that can run at once (parallel investigation, multi-file changes, fan-out research, multi-lane review). Uses native Copilot task/fleet scheduling and integrates worker results skeptically.
| name | csw-review |
| description | Use to run a rigorous multi-lane review before declaring work done — for any broad, risky, shared, security-sensitive, or release-facing change. Uses native scheduling for independent review lanes and gates on ALL of them passing with a single binding verdict. |
Run independent review lanes in parallel and accept the work only if EVERY lane passes. One failing lane = overall REJECTION.
Use native task subagents to run these as independent workers, routing each to the
verifier agent with a lane-specific brief. Use /fleet for user-visible parallel
execution and /tasks to inspect or cancel work. Enforce non-mutation with host
deny/available-tool policy before launch; a review prompt is not a security boundary.
Each lane's worker prompt must be self-contained: include the goal, the diff (or how to obtain it), the success criteria, and the evidence locations.
Worker verdicts are claims. Before accepting an APPROVE, spot-check it yourself:
re-read the diff, re-run the cited tests, open the cited evidence artifacts.
Replay command-backed criteria with csw-runtime verify --id <C0NN> -- <argv...>;
bind file-backed real-surface evidence with csw-runtime artifact --id <C0NN> --path <workspace-file> --summary "<observed outcome>". Free-text evidence cannot
pass a criterion.
Combine the lanes:
Record the result. If a goal runtime is active, completion still requires
csw-runtime complete to pass (criteria + machine receipts + zero blockers) — this review
is the human-facing quality gate layered on top of the oracle, not a replacement
for it. Unresolved review findings should be filed as blockers
(csw-runtime blocker add) so they block completion until addressed.
