feat: Payment Compliance Pack — Semgrep + CodeQL + State Machine + Schema

Ausfuehrbares Pruefpaket fuer Payment-Terminal-Systeme:

1. Semgrep-Regeln (25 Regeln in 5 Dateien):
   - Logging: Sensitive Daten, Tokens, Debug-Flags
   - Crypto: MD5/SHA1/DES/ECB, Hardcoded Secrets, Weak Random, TLS
   - API: Debug-Routes, Exception Leaks, IDOR, Input Validation
   - Config: Test-Endpoints, CORS, Cookies, Retry
   - Data: Telemetrie, Cache, Export, Queue, Testdaten

2. CodeQL Query-Specs (5 Briefings):
   - Sensitive Data → Logs
   - Sensitive Data → HTTP Response
   - Tenant Context Loss
   - Sensitive Data → Telemetry
   - Cache/Export Leak

3. State-Machine-Tests (10 Testfaelle):
   - 11 Zustaende, 15 Events, 8 Invarianten
   - Duplicate Response, Timeout+Late Success, Decline
   - Invalid Reversal, Cancel, Backend Timeout
   - Parallel Reversal, Unknown Response, Reconnect
   - Late Response after Cancel

4. Finding Schema (JSON Schema):
   - Einheitliches Format fuer alle Engines
   - control_id, engine, status, confidence, evidence, verdict_text

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

ai-compliance-sdk/payment-compliance-pack/codeql/sensitive-data-to-response.md

@@ -0,0 +1,19 @@
+# CodeQL Query: Sensitive Data to HTTP Response
+
+## Ziel
+Finde Fluesse sensibler Daten in HTTP-/API-Responses oder Exception-Bodies.
+
+## Sources
+- Sensible Payment-Felder: pan, cvv, track2, cardNumber, pin, expiry
+- Interne Payment DTOs mit sensitiven Attributen
+
+## Sinks
+- JSON serializer / response builder
+- Exception payload / error handler response
+- Template rendering output
+
+## Mapped Controls
+- `API-009`: API-Antworten minimieren sensible Daten
+- `API-015`: Interne Fehler ohne sensitive Daten an Client
+- `ERROR-005`: Ausnahmebehandlung gibt keine sensitiven Rohdaten zurueck
+- `REPORT-006`: Reports offenbaren nur rollenerforderliche Daten