Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →ECC用の証拠ベースの自動化インベントリとオーバーラップ監査ワークフロー。ユーザーがどのジョブ、フック、コネクタ、MCPサーバー、またはラッパーがライブか、壊れているか、冗長であるか、修正前に不足しているかを知りたい場合に使用します。
.claude/skills/affaan-m-automation-audit-ops/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-13 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 70% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 3% | 0% |
当用户询问哪些自动化正在运行、哪些任务出现故障、哪里存在重叠，或者哪些工具和连接器当前正在实际发挥作用时，请使用此技能。
这是一项以审计为先的操作技能。其任务是在重写任何内容之前，生成一份有证据支持的清单以及一套保留/合并/删除/下一步修复的建议集。
在相关时，将这些 ECC 原生技能引入工作流程：
workspace-surface-audit 用于连接器、MCP、钩子和应用清单knowledge-ops 当审计需要将实时仓库的真实情况与持久上下文进行核对时github-ops 当答案依赖于 CI、计划工作流、议题或 PR 自动化时ecc-tools-cost-audit 当真正的问题是兄弟应用仓库中的 webhook 扇出、队列任务或计费消耗时research-ops 当需要将本地清单与当前平台支持或公开文档进行比较时verification-loop 用于证明修复后的状态，而不是依赖假设的恢复在理论化之前，先读取当前的实时表面：
按表面分组：
对于每个发现的自动化，标记：
然后对问题类型进行分类：
为每个重要声明提供具体来源：
如果当前状态不明确，请直接说明，而不是假装审计已完成。
对于每个重叠或可疑的表面，返回一个决策：
其价值在于将杂乱的自动化整合到一条规范的 ECC 路径中，而不是保留每一条历史路径。
text当前表面 - 自动化 - 来源 - 实时状态 - 证据 发现 - 活跃故障 - 重叠 - 过时状态 - 缺失能力 建议 - 保留 - 合并 - 删除 - 下次修复 下一步ECC行动 - 需加强的具体技能/钩子/工作流/应用通道
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 20,357 | 6,425 | -68% | 1 | 1 | 0% | 3,249 | 1,536 | -53% | 0 | 0 | — |
case-02 | fail→fail | 18,719 | 6,469 | -65% | 1 | 1 | 0% | 2,950 | 1,394 | -53% | 0 | 0 | — |
case-13 | fail→pass | 9,523 | 4,781 | -50% | 1 | 1 | 0% | 1,557 | 1,728 | +11% | 0 | 0 | — |
case-03 | fail→pass | 14,530 | 7,502 | -48% | 1 | 1 | 0% | 2,491 | 1,493 | -40% | 0 | 0 | — |
case-04 | pass→pass | 17,109 | 6,710 | -61% | 1 | 1 | 0% | 1,645 | 2,023 | +23% | 0 | 0 | — |
case-05 | fail→pass | 13,294 | 7,824 | -41% | 1 | 1 | 0% | 2,026 | 2,275 | +12% | 0 | 0 | — |
case-06 | fail→pass | 5,488 | 3,120 | -43% | 1 | 1 | 0% | 873 | 1,484 | +70% | 0 | 0 | — |
case-07 | fail→pass | 10,286 | 4,830 | -53% | 1 | 1 | 0% | 1,708 | 1,766 | +3% | 0 | 0 | — |
case-08 | fail→pass | 9,055 | 4,047 | -55% | 1 | 1 | 0% | 1,482 | 1,703 | +15% | 0 | 0 | — |
case-09 | fail→pass | 15,004 | 14,285 | -5% | 1 | 1 | 0% | 2,286 | 3,353 | +47% | 0 | 0 | — |
case-10 | pass→pass | 13,540 | 7,896 | -42% | 1 | 1 | 0% | 2,049 | 2,201 | +7% | 0 | 0 | — |
case-11 | fail→pass | 7,759 | 7,452 | -4% | 1 | 1 | 0% | 1,298 | 2,197 | +69% | 0 | 0 | — |
case-12 | fail→pass | 10,592 | 6,738 | -36% | 1 | 1 | 0% | 1,769 | 1,978 | +12% | 0 | 0 | — |
case-14 | fail→pass | 14,215 | 9,416 | -34% | 1 | 1 | 0% | 2,054 | 2,489 | +21% | 0 | 0 | — |
case-15 | fail→pass | 10,944 | 9,615 | -12% | 1 | 1 | 0% | 1,661 | 2,025 | +22% | 0 | 0 | — |
case-16 | fail→pass | 15,627 | 8,452 | -46% | 1 | 1 | 0% | 2,289 | 2,324 | +2% | 0 | 0 | — |
case-17 | fail→pass | 13,870 | 3,477 | -75% | 1 | 1 | 0% | 1,963 | 1,512 | -23% | 0 | 0 | — |
case-18 | fail→pass | 16,379 | 4,659 | -72% | 1 | 1 | 0% | 2,260 | 1,712 | -24% | 0 | 0 | — |
case-19 | fail→pass | 13,545 | 8,232 | -39% | 1 | 1 | 0% | 2,095 | 2,166 | +3% | 0 | 0 | — |
case-20 | pass→fail | 9,720 | 17,894 | +84% | 1 | 1 | 0% | 1,540 | 3,931 | +155% | 0 | 0 | — |
case-21 | pass→pass | 10,066 | 12,434 | +24% | 1 | 1 | 0% | 2,042 | 3,349 | +64% | 0 | 0 | — |
case-22 | pass→pass | 12,084 | 11,994 | -1% | 1 | 1 | 0% | 2,072 | 3,049 | +47% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +64 percentage points is the difference between those two pass rates over the 20 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.