Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →用于合同条款风险审查，基于关键词与规则识别高风险表述（单方决定、责任失衡、模糊期限、无限责任等）并输出修订建议。适用于法务风控预审、合同谈判前风险筛查场景。
.claude/skills/aifinlab-clause-risk-review-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | -53% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 50% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -19% | 0% |
本技能用于识别合同条款中的常见风险表达，输出风险等级与修订建议，辅助法务快速聚焦重点条款。
支持 JSON 数组或 JSONL，每条记录建议包含：
contract_nameclause_noclause_titleclause_text使用 scripts/clause_risk_review.py 生成报告：
bashpython scripts/clause_risk_review.py \ --input clauses.json \ --output clause_risk_report.md
可选参数：
--rules：自定义风险规则 JSON--top：展示高风险条款数量（默认 20）| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-08 | fail→pass | 7,841 | 1,783 | -77% | 1 | 1 | 0% | 1,256 | 592 | -53% | 0 | 0 | — |
case-01 | fail→fail | 10,030 | 27,674 | +176% | 1 | 1 | 0% | 1,665 | 4,342 | +161% | 0 | 0 | — |
case-02 | fail→fail | 15,176 | 28,638 | +89% | 1 | 1 | 0% | 2,214 | 4,896 | +121% | 0 | 0 | — |
case-03 | fail→pass | 20,790 | 27,862 | +34% | 1 | 1 | 0% | 3,048 | 4,570 | +50% | 0 | 0 | — |
case-04 | pass→pass | 15,526 | 5,378 | -65% | 1 | 1 | 0% | 2,823 | 1,315 | -53% | 0 | 0 | — |
case-05 | fail→fail | 16,824 | 12,402 | -26% | 1 | 1 | 0% | 2,829 | 2,449 | -13% | 0 | 0 | — |
case-06 | fail→pass | 20,331 | 7,882 | -61% | 1 | 1 | 0% | 3,019 | 1,462 | -52% | 0 | 0 | — |
case-07 | fail→pass | 20,085 | 18,650 | -7% | 1 | 1 | 0% | 2,882 | 3,099 | +8% | 0 | 0 | — |
case-09 | fail→pass | 20,065 | 14,149 | -29% | 1 | 1 | 0% | 3,177 | 2,559 | -19% | 0 | 0 | — |
case-10 | pass→pass | 18,062 | 11,315 | -37% | 1 | 1 | 0% | 2,590 | 2,104 | -19% | 0 | 0 | — |
case-11 | pass→pass | 16,282 | 14,415 | -11% | 1 | 1 | 0% | 2,428 | 2,580 | +6% | 0 | 0 | — |
case-12 | fail→pass | 8,861 | 2,740 | -69% | 1 | 1 | 0% | 1,530 | 795 | -48% | 0 | 0 | — |
case-13 | fail→fail | 27,290 | 26,448 | -3% | 1 | 1 | 0% | 4,436 | 4,517 | +2% | 0 | 0 | — |
case-14 | fail→pass | 22,024 | 19,635 | -11% | 1 | 1 | 0% | 3,059 | 3,229 | +6% | 0 | 0 | — |
case-15 | pass→pass | 8,349 | 8,902 | +7% | 1 | 1 | 0% | 1,291 | 1,571 | +22% | 0 | 0 | — |
case-16 | fail→pass | 10,864 | 2,523 | -77% | 1 | 1 | 0% | 1,927 | 777 | -60% | 0 | 0 | — |
case-17 | fail→pass | 9,458 | 3,896 | -59% | 1 | 1 | 0% | 1,517 | 939 | -38% | 0 | 0 | — |
case-18 | fail→pass | 10,486 | 3,430 | -67% | 1 | 1 | 0% | 1,814 | 823 | -55% | 0 | 0 | — |
case-19 | fail→pass | 16,869 | 5,343 | -68% | 1 | 1 | 0% | 2,653 | 1,189 | -55% | 0 | 0 | — |
case-20 | pass→pass | 15,081 | 16,732 | +11% | 1 | 1 | 0% | 2,313 | 2,854 | +23% | 0 | 0 | — |
case-21 | pass→pass | 14,933 | 12,221 | -18% | 1 | 1 | 0% | 2,140 | 2,072 | -3% | 0 | 0 | — |
case-22 | pass→pass | 17,704 | 7,149 | -60% | 1 | 1 | 0% | 2,423 | 1,475 | -39% | 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. The headline lift of +50 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.