Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →アクティブなリポジトリ、MCPサーバー、プラグイン、コネクター、環境サーフェス、ツールのセットアップを監査し、最も価値の高いECCネイティブスキル、フック、エージェント、オペレーターワークフローを推奨する。ユーザーがClaude Codeのセットアップを支援してほしい場合や、環境で実際に何が使えるかを理解したい場合に使用する。
.claude/skills/affaan-m-workspace-surface-audit/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-14 | ✓→✗ | ▼ Worse | 422% | 0% |
| case-18 | ✓→✗ | ▼ Worse | -18% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | 17% | 0% |
只读审计技能，用于回答"这个工作区和机器当前实际上能做什么，以及我们下一步应该添加或启用什么？"
这是 ECC 原生对设置审计插件的回答。除非用户明确要求后续实现，否则不会修改文件。
.env、.mcp.json、插件设置或连接的应用表面，以发现缺失的工作流层仅检查回答该问题所需的文件和设置：
package.json、锁定文件、语言标记、框架配置、README.md.mcp.json、.lsp.json、.claude/settings*.json、.codex/*AGENTS.md、CLAUDE.md、安装清单、钩子配置.env* 文件STRIPE_API_KEY、TWILIO_AUTH_TOKEN、FAL_KEY生成简洁的清单：
如果某个表面仅以原始形式存在，请指出。例如：
将工作区与以下内容进行比较：
不要仅列出名称。对于每个比较，回答：
对于每个实际差距，推荐正确的 ECC 原生形态：
| 差距类型 | 首选 ECC 形态 | |----------|---------------------| | 可重复的操作工作流 | 技能 | | 自动执行或副作用 | 钩子 | | 专门的委派角色 | 代理 | | 外部工具桥接 | MCP 服务器或连接器 | | 安装/引导指南 | 设置或审计技能 |
当需求是操作性的而非基础设施性的时，默认使用面向用户的技能来编排现有工具。
按此顺序返回五个部分：
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 8,283 | 6,622 | -20% | 1 | 1 | 0% | 644 | 1,536 | +139% | 0 | 0 | — |
case-02 | fail→fail | 12,613 | 4,622 | -63% | 1 | 1 | 0% | 2,164 | 1,589 | -27% | 0 | 0 | — |
case-03 | fail→fail | 6,163 | 4,884 | -21% | 1 | 1 | 0% | 591 | 1,500 | +154% | 0 | 0 | — |
case-04 | fail→fail | 6,401 | 4,427 | -31% | 1 | 1 | 0% | 1,208 | 1,471 | +22% | 0 | 0 | — |
case-05 | fail→fail | 9,311 | 6,023 | -35% | 1 | 1 | 0% | 1,776 | 1,471 | -17% | 0 | 0 | — |
case-06 | pass→pass | 11,189 | 6,102 | -45% | 1 | 1 | 0% | 1,855 | 2,175 | +17% | 0 | 0 | — |
case-07 | pass→pass | 7,085 | 5,834 | -18% | 1 | 1 | 0% | 1,221 | 2,239 | +83% | 0 | 0 | — |
case-08 | pass→pass | 11,229 | 8,453 | -25% | 1 | 1 | 0% | 1,939 | 2,545 | +31% | 0 | 0 | — |
case-09 | fail→pass | 13,727 | 4,958 | -64% | 1 | 1 | 0% | 2,472 | 2,003 | -19% | 0 | 0 | — |
case-10 | pass→pass | 11,123 | 5,009 | -55% | 1 | 1 | 0% | 1,918 | 1,997 | +4% | 0 | 0 | — |
case-11 | fail→fail | 22,582 | 6,239 | -72% | 1 | 1 | 0% | 3,878 | 1,620 | -58% | 0 | 0 | — |
case-12 | pass→pass | 11,491 | 8,363 | -27% | 1 | 1 | 0% | 1,956 | 2,529 | +29% | 0 | 0 | — |
case-13 | pass→pass | 9,478 | 5,949 | -37% | 1 | 1 | 0% | 1,603 | 2,173 | +36% | 0 | 0 | — |
case-14 | pass→fail | 3,150 | 4,976 | +58% | 1 | 1 | 0% | 297 | 1,549 | +422% | 0 | 0 | — |
case-15 | fail→pass | 9,128 | 3,582 | -61% | 1 | 1 | 0% | 1,632 | 1,877 | +15% | 0 | 0 | — |
case-16 | pass→pass | 12,500 | 6,586 | -47% | 1 | 1 | 0% | 2,789 | 2,444 | -12% | 0 | 0 | — |
case-17 | pass→pass | 10,996 | 3,722 | -66% | 1 | 1 | 0% | 2,058 | 1,822 | -11% | 0 | 0 | — |
case-18 | pass→fail | 11,376 | 3,246 | -71% | 1 | 1 | 0% | 2,003 | 1,646 | -18% | 0 | 0 | — |
case-19 | pass→pass | 7,609 | 3,647 | -52% | 1 | 1 | 0% | 1,332 | 1,674 | +26% | 0 | 0 | — |
case-20 | pass→pass | 4,522 | 7,775 | +72% | 1 | 1 | 0% | 921 | 2,717 | +195% | 0 | 0 | — |
case-21 | pass→pass | 3,607 | 8,084 | +124% | 1 | 1 | 0% | 679 | 2,612 | +285% | 0 | 0 | — |
case-22 | pass→pass | 19,433 | 12,662 | -35% | 1 | 1 | 0% | 4,636 | 4,382 | -5% | 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 15 counted toward the lift figure. The other 7 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 0 percentage points is the difference between those two pass rates over the 15 comparable cases. 4 cases got worse with the skill loaded, and they are 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.