Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Anti-Money Laundering (AML) and Know Your Customer (KYC) compliance workflow. Sanctions screening, PEP detection, transaction monitoring, suspicious activity reporting (SAR), and OFAC compliance.
.claude/skills/mkurman-aml-compliance/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-21 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -32% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -16% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -27% | 0% |
AML/KYC compliance covers sanctions screening, PEP detection, transaction monitoring, currency transaction reports (CTR), suspicious activity reports (SAR), and OFAC compliance. Essential for fintech, banking, and payment applications handling regulated financial transactions.
bashuv pip install requests # for sanctions API integration
pythonTHRESHOLDS = {"ctr": 10000, "structuring_lookback": 5000} HIGH_RISK_COUNTRIES = {"IR", "KP", "SY", "CU", "MM"} def screen_tx(tx): alerts = [] if tx.amount >= THRESHOLDS["ctr"]: alerts.append("CTR required — cash transaction over $10k") if tx.country in HIGH_RISK_COUNTRIES: alerts.append("OFAC sanctioned jurisdiction — enhanced due diligence") if tx.is_pep: alerts.append("PEP flagged — enhanced monitoring") return alerts
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-21 | fail→pass | 6,211 | 2,027 | -67% | 1 | 1 | 0% | 992 | 595 | -40% | 0 | 0 | — |
case-01 | pass→pass | 11,553 | 8,977 | -22% | 1 | 1 | 0% | 1,769 | 1,886 | +7% | 0 | 0 | — |
case-02 | fail→pass | 21,965 | 6,418 | -71% | 1 | 1 | 0% | 1,910 | 1,307 | -32% | 0 | 0 | — |
case-03 | fail→pass | 6,461 | 3,928 | -39% | 1 | 1 | 0% | 948 | 798 | -16% | 0 | 0 | — |
case-04 | pass→pass | 18,048 | 2,665 | -85% | 1 | 1 | 0% | 1,431 | 617 | -57% | 0 | 0 | — |
case-05 | pass→pass | 11,180 | 3,933 | -65% | 1 | 1 | 0% | 1,659 | 995 | -40% | 0 | 0 | — |
case-06 | pass→pass | 12,838 | 9,636 | -25% | 1 | 1 | 0% | 2,021 | 1,650 | -18% | 0 | 0 | — |
case-07 | fail→pass | 7,861 | 3,534 | -55% | 1 | 1 | 0% | 1,290 | 469 | -64% | 0 | 0 | — |
case-08 | fail→pass | 13,661 | 1,823 | -87% | 1 | 1 | 0% | 703 | 512 | -27% | 0 | 0 | — |
case-09 | pass→pass | 7,951 | 2,365 | -70% | 1 | 1 | 0% | 1,238 | 683 | -45% | 0 | 0 | — |
case-10 | pass→pass | 16,788 | 2,206 | -87% | 1 | 1 | 0% | 1,786 | 684 | -62% | 0 | 0 | — |
case-11 | pass→pass | 8,896 | 4,524 | -49% | 1 | 1 | 0% | 1,403 | 1,079 | -23% | 0 | 0 | — |
case-12 | fail→pass | 11,143 | 3,533 | -68% | 1 | 1 | 0% | 1,719 | 886 | -48% | 0 | 0 | — |
case-13 | fail→pass | 13,596 | 3,580 | -74% | 1 | 1 | 0% | 1,642 | 570 | -65% | 0 | 0 | — |
case-14 | fail→pass | 4,036 | 3,122 | -23% | 1 | 1 | 0% | 519 | 783 | +51% | 0 | 0 | — |
case-15 | fail→pass | 9,887 | 1,636 | -83% | 1 | 1 | 0% | 1,655 | 475 | -71% | 0 | 0 | — |
case-16 | pass→pass | 4,464 | 1,992 | -55% | 1 | 1 | 0% | 697 | 539 | -23% | 0 | 0 | — |
case-17 | pass→pass | 2,815 | 2,889 | +3% | 1 | 1 | 0% | 402 | 506 | +26% | 0 | 0 | — |
case-18 | pass→pass | 4,981 | 2,986 | -40% | 1 | 1 | 0% | 786 | 621 | -21% | 0 | 0 | — |
case-19 | fail→pass | 15,561 | 5,163 | -67% | 1 | 1 | 0% | 3,157 | 1,299 | -59% | 0 | 0 | — |
case-20 | pass→pass | 11,409 | 3,371 | -70% | 1 | 1 | 0% | 1,521 | 535 | -65% | 0 | 0 | — |
case-22 | pass→pass | 15,615 | 13,676 | -12% | 1 | 1 | 0% | 2,489 | 2,365 | -5% | 0 | 0 | — |
case-23 | pass→pass | 4,953 | 3,953 | -20% | 1 | 1 | 0% | 792 | 983 | +24% | 0 | 0 | — |
case-24 | pass→pass | 17,552 | 19,846 | +13% | 1 | 1 | 0% | 2,860 | 3,760 | +31% | 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. 24 cases were attempted, and 23 counted toward the lift figure. The other 1 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 +42 percentage points is the difference between those two pass rates over the 23 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.