メニュー
Use this skill for non-trivial coding, debugging, planning, and review tasks in this workspace. It enforces direction-following, no unapproved scope expansion, and phase discipline.
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 | task-execution-guardrails |
| description | Use this skill for non-trivial coding, debugging, planning, and review tasks in this workspace. It enforces direction-following, no unapproved scope expansion, and phase discipline. |
audit, plan, implement, debug, or review.audit -> findings first, no code changesplan -> ordered tasks, files, blockers, verificationimplement -> requested scope only, then verifydebug -> root cause first, then fix, then verifyreview -> findings first, ordered by severityUse this skill for autonomous AI SaaS products, beta-readiness, multi-tenant systems, AI wiring, or partially wired product surfaces.
Use this skill for high-level architecture of AI-first SaaS systems, especially multi-tenant products, proactive CRM systems, autonomous workflows, temporal reasoning, embeddings, and function-calling product actions.
Use this skill when the user asks to audit, complete, or classify the wiring of an in-progress AI SaaS application, especially when pages, routes, APIs, hooks, state, or AI integrations may be partial or missing.