| name | regulatory-mapping |
| description | Ulpian's method — obligation catalogs, obligation→control→evidence mapping, gap prioritization, the EU AI Act classification tree, MiCA/MiCAR orientation, sectoral regime index (SOX/HIPAA/FCPA), sanctions/export trigger checklists, and horizon scanning. |
| allowed-tools | ["Read","Write","Glob","Grep"] |
Regulatory Mapping
Obligation Catalog Format
For each applicable regime, catalog at article level:
| Regime | Provision | Obligation (verb + object) | Addressee | Trigger | Deadline | Penalty exposure |
Rules: obligations are quoted or tightly paraphrased with pin cites
(Tribonian tags currency); conditional obligations carry their trigger
explicitly; "best efforts"-grade duties are flagged as judgment-dependent.
Obligation → Control → Evidence Mapping
| Obligation (ref) | Control that satisfies it | Control owner | Evidence | Status |
The Ulpian rule: the control column names a control that EXISTS.
Status ∈ mapped / partial / GAP / [CONTROL UNKNOWN — verify].
Inventing a control to complete a row is the regulatory equivalent of a
fabricated citation.
Gap Prioritization
Score each gap: enforcement probability (30%) × penalty severity incl.
personal/criminal exposure (25%) × remediation cost+time (20%) × detection
likelihood (15%) × reputational amplification (10%).
≥3.5 remediate-now · 2.5–3.4 quarter-plan · <2.5 backlog-with-owner.
Feeds compliance-coverage@1.
EU AI Act Classification Tree
Is it an AI system (Art. 3(1): machine-based, infers from input, generates outputs)?
├── No → out of scope (document why)
└── Yes
├── Prohibited practice (Art. 5: manipulation, social scoring,
│ untargeted scraping, real-time remote biometric ID…)? → STOP + veto
├── High-risk (Art. 6 + Annex III: employment, credit, essential
│ services, law enforcement…, or Annex I product-safety component)?
│ → Ch. III duties: risk-mgmt system, data governance, technical
│ docs, logging, transparency, human oversight, accuracy/robustness;
│ deployer duties distinct from provider duties — classify the role
├── GPAI model? → Ch. V: technical docs, copyright policy, training-data
│ summary; +systemic-risk duties above compute threshold
├── Limited-risk (chatbots, synthetic content)? → Art. 50 transparency
└── Minimal → voluntary codes
Always record: role (provider/deployer/importer/distributor), tier,
applicable duty set, and applicability dates (the Act phases in — temporal
tags mandatory). Feeds eu-ai-act-classification@1.
MiCA/MiCAR Orientation (crypto-assets, EU)
Token taxonomy first: ART (asset-referenced) / EMT (e-money) / other
crypto-asset (utility) — each with its own whitepaper, authorization, and
conduct duties; CASP licensing for service providers; market-abuse rules
across the board. Non-compliant stablecoin delisting precedent noted.
Detailed analysis routes to the regime catalog with current-text
verification.
Sectoral Index (route-to-depth markers)
- SOX — ICFR, §302/906 certifications, auditor independence; route
control testing to CCO surface.
- HIPAA — PHI: covered-entity/business-associate analysis, Security
Rule safeguards, BAA paper (with Angerona).
- FCPA / UKBA — anti-bribery: third-party diligence, facilitation
payments (banned under UKBA), books-and-records.
- Sanctions/Export (Janus's checklists):
- Screening protocol: counterparty → beneficial ownership (50% rule) →
destination → end-use red flags (evasive answers, freight-forwarder
end users, transshipment hubs, cash above market).
- Classification triggers: encryption, dual-use Annex I categories,
AI/compute thresholds, military/space nexus.
- Output defines the screening to run against CURRENT lists — never
certifies clearance from memory (Phase-1 limits).
Horizon Scanning Discipline
Each regulatory matter notes: pending amendments to applied regimes,
consultation windows open, enforcement-trend signals, and a
re-assessment trigger date. Unknown ⇒ [HORIZON UNSCANNED] — silence is
not stability.
Required Logging
Regime list with applicability reasoning · catalog coverage (articles
mapped vs. total) · gap register with scores · as-of statement for every
regime text relied on.