Part of the Overpowered skill suite. Require evidence proportional to material claims and clearly separate evidence, inference, and uncertainty. Use for research, analysis, recommendations, reports, audits, or decisions where unsupported confidence would be harmful; do not use for purely creative or stylistic tasks.
安全
completion-audit
试用Part of the Overpowered skill suite. Independently determine whether a claimed task outcome is actually complete by reconstructing completion criteria and checking fresh evidence. Use before declaring work done, accepted, migrated, reconciled, fixed, or delivered; not as a general planning loop.
它能做什么
Part of the Overpowered skill suite. Independently determine whether a claimed task outcome is actually complete by reconstructing completion criteria and checking fresh evidence. Use before declaring work done, accepted, migrated, reconciled, fixed, or delivered; not as a general planning loop.
技能文档
Completion Audit
Core rule
Activity is not completion. A completion claim is valid only when its material success criteria are supported by evidence.
Procedure
- Restate the claimed outcome without trusting the worker's summary.
- Derive observable completion criteria from the original objective, constraints, and acceptance conditions.
- Identify the authoritative evidence for each criterion.
- Inspect or run fresh checks when tools permit.
- Grade each criterion:
- PROVEN — sufficient evidence;
- NOT PROVEN — evidence contradicts the claim or is missing;
- NOT APPLICABLE — criterion does not apply, with reason.
- Return the actual status. Do not soften
NOT PROVENinto “probably done.”
Output contract
Claimed outcome: ...
Criterion | Evidence | Status
...
Verdict: PROVEN / NOT PROVEN
Remaining work or missing evidence: ...
Independence rule
Prefer source artifacts, test outputs, database checks, logs, or direct observations over the executing agent's narrative.
Gotchas
- Do not add arbitrary criteria that were never part of the objective or necessary invariants.
- A generated file existing is not proof that its contents are correct.
- A successful tool call is not proof that the desired business outcome occurred.
- If verification is impossible, say
NOT PROVEN, notFAILED, unless evidence shows failure.
Stop condition
Stop when every material criterion has an explicit status and the overall verdict follows from those statuses.
相关技能
Audit another Codex skill for structural compliance, trigger quality, instruction clarity, reuse of scripts or references, and overall maintainability. Use w...
16 次安装
Part of the Overpowered skill suite. Expose material assumptions, hidden premises, and unresolved unknowns in a plan, analysis, design, or decision. Use before committing to work when an incorrect assumption could change the approach; avoid turning trivial low-impact unknowns into bureaucracy.
Part of the Overpowered skill suite. Route knowledge-work and enterprise-automation tasks through the smallest relevant combination of Overpowered skills, and invoke gear-up only when a material execution capability is genuinely missing. Use when multiple suite skills may apply or when starting a non-trivial task involving organizational knowledge, conflicting evidence, evolving artifacts, business rules, process automation, risky side effects, verifiable completion, or an uncovered capability gap.
Part of the Overpowered skill suite. Create a durable, self-contained state package so a fresh agent, session, or different harness can resume non-trivial work without the original chat history. Use for long-running work, context resets, handoffs, or pauses; not for simple finished tasks.
Part of the Overpowered skill suite. Transform a described human/business process into an executable automation design by separating deterministic steps, agentic reasoning, existing systems, and human gates; define triggers, data, decisions, exceptions, controls, and verification. Use when the goal is to automate a process, not merely understand it or interview the user about it.