Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Strategic advisory for fintech founders on US/EU regulatory triggers, license-vs-partner, KYC/AML, and embedded finance. Use when scoping a fintech idea or regulatory exposure, or mentioning fintech, money transmitter, neobank, KYC, or PSD2.
.claude/skills/borghei-fintech-advisor/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 137% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -4% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 37% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-22 | ✗→✓ | ▲ Improved | 33% | 0% |
Strategic frameworks for fintech founders, operators, and product leaders. Knowledge-heavy by design — the right answer in fintech is usually a regulatory and economic judgment, not a calculation.
> Disclaimer: This skill provides frameworks and orientation. It is not legal, regulatory, securities, tax, or investment advice. Every fintech business needs licensed legal counsel. Use this skill to organize internal thinking; engage specialist counsel for binding decisions.
fintech, payments, banking, neobank, lending, money transmitter, KYC, AML, PSD2, open banking, BaaS, banking-as-a-service, embedded finance, card issuing, ACH, SEPA, stablecoin, crypto, broker-dealer, RIA, regulation, compliance
python scripts/regulatory_trigger_checker.py business_description.txtreferences/license_vs_partner_playbook.mdGoal: Understand which US / EU regulatory regimes a proposed fintech business model triggers, before committing to architecture.
Steps:
Time Estimate: 4-8 weeks of legal scoping for a meaningful new build.
Goal: Decide whether to get the regulated capability yourself, or buy it from a partner.
Steps:
license_vs_partner_playbook.md: cost, time, control, economicsTime Estimate: 6-12 weeks for major capability decisions.
Goal: Build a KYC/AML program that satisfies regulators and doesn't kill conversion.
Steps:
references/kyc_aml_basics.mdTime Estimate: 8-16 weeks for first-time program design.
Scans a business description for keywords and patterns that map to regulatory regimes in the US and EU. Output is a list of candidate triggers, not a legal opinion.
bashpython scripts/regulatory_trigger_checker.py business_description.txt python scripts/regulatory_trigger_checker.py business_description.txt --json
Triggers detected:
references/regulatory_landscape.md — Map of US and EU fintech regulators, what each covers, common trigger patternsreferences/license_vs_partner_playbook.md — When to get a license, when to partner, partner failure planningreferences/kyc_aml_basics.md — KYC tiers, risk-based monitoring, MLRO role, common pitfallsreferences/embedded_finance_patterns.md — BaaS architecture, distribution-led fintech, B2B2C patternsassets/regulatory_architecture_template.md — Document template for capturing regulatory decisions and partner choicesc-level-advisor/cs-fundraising-advisor — investors expect a clear regulatory architectureengineering/cs-security-engineer — fintech security goes beyond standard SaaSlegal/ skills for contract / partner agreementsbusiness-growth/pricing-strategy — fintech pricing has unusual constraints (interchange, FX spread, float)| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 42,197 | 28,510 | -32% | 1 | 1 | 0% | 6,237 | 5,759 | -8% | 0 | 0 | — |
case-02 | pass→pass | 15,542 | 19,515 | +26% | 1 | 1 | 0% | 2,566 | 4,341 | +69% | 0 | 0 | — |
case-03 | fail→fail | 22,190 | 17,036 | -23% | 1 | 1 | 0% | 3,630 | 3,870 | +7% | 0 | 0 | — |
case-04 | fail→pass | 10,395 | 13,465 | +30% | 1 | 1 | 0% | 1,473 | 3,496 | +137% | 0 | 0 | — |
case-05 | fail→fail | 12,528 | 3,711 | -70% | 1 | 1 | 0% | 1,964 | 2,169 | +10% | 0 | 0 | — |
case-06 | fail→fail | 18,066 | 14,688 | -19% | 1 | 1 | 0% | 2,557 | 3,633 | +42% | 0 | 0 | — |
case-07 | fail→fail | 21,092 | 17,759 | -16% | 1 | 1 | 0% | 2,963 | 3,884 | +31% | 0 | 0 | — |
case-08 | pass→pass | 15,260 | 16,690 | +9% | 1 | 1 | 0% | 2,261 | 3,836 | +70% | 0 | 0 | — |
case-09 | fail→fail | 16,731 | 12,571 | -25% | 1 | 1 | 0% | 2,455 | 3,286 | +34% | 0 | 0 | — |
case-10 | fail→fail | 11,729 | 9,638 | -18% | 1 | 1 | 0% | 1,762 | 2,927 | +66% | 0 | 0 | — |
case-11 | pass→pass | 18,527 | 19,534 | +5% | 1 | 1 | 0% | 2,865 | 4,600 | +61% | 0 | 0 | — |
case-12 | pass→pass | 17,427 | 12,733 | -27% | 1 | 1 | 0% | 2,587 | 3,355 | +30% | 0 | 0 | — |
case-13 | pass→pass | 18,688 | 23,999 | +28% | 1 | 1 | 0% | 2,913 | 5,233 | +80% | 0 | 0 | — |
case-14 | pass→pass | 11,600 | 11,907 | +3% | 1 | 1 | 0% | 1,958 | 3,324 | +70% | 0 | 0 | — |
case-15 | fail→pass | 12,628 | 3,186 | -75% | 1 | 1 | 0% | 1,933 | 1,854 | -4% | 0 | 0 | — |
case-16 | fail→pass | 9,458 | 2,689 | -72% | 1 | 1 | 0% | 1,367 | 1,879 | +37% | 0 | 0 | — |
case-17 | fail→pass | 20,977 | 2,889 | -86% | 1 | 1 | 0% | 3,249 | 1,894 | -42% | 0 | 0 | — |
case-18 | pass→pass | 16,562 | 13,843 | -16% | 1 | 1 | 0% | 2,372 | 3,443 | +45% | 0 | 0 | — |
case-19 | pass→fail | 16,993 | 18,389 | +8% | 1 | 1 | 0% | 2,475 | 3,930 | +59% | 0 | 0 | — |
case-20 | pass→pass | 8,593 | 10,351 | +20% | 1 | 1 | 0% | 1,334 | 3,112 | +133% | 0 | 0 | — |
case-21 | pass→pass | 6,370 | 9,589 | +51% | 1 | 1 | 0% | 981 | 2,856 | +191% | 0 | 0 | — |
case-22 | fail→pass | 17,616 | 11,496 | -35% | 1 | 1 | 0% | 2,483 | 3,298 | +33% | 0 | 0 | — |
case-23 | fail→fail | 16,474 | 16,013 | -3% | 1 | 1 | 0% | 2,332 | 3,503 | +50% | 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. 23 cases were attempted. The headline lift of +17 percentage points is the difference between those two pass rates over the 23 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.