Commit Graph

6 Commits

Author SHA1 Message Date
Benjamin Admin
a7c6ffe4dd feat(control-pipeline): add SDK endpoint demo package for applicability tests
Request payloads + response contract + api_runner.py for 6 priority cases.
Can be run directly against /v1/applicability/evaluate endpoint.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-23 19:11:44 +02:00
Benjamin Admin
ae5c5c24eb feat(control-pipeline): add applicability demo test package with evaluator
6 priority demo cases with golden outputs, evaluator.py and run_demo.py:
- CASE-001: Webshop+Stripe (anti-PSD2 false positive)
- CASE-002: Bank+TAN-Generator (scope override for batteries)
- CASE-004: FinTech Wallet (true positive PSD2/AML)
- CASE-006: SaaS+SMS Gateway (anti-TKG false positive)
- CASE-008: Software→IoT Hardware (multi-regime scope)
- CASE-011: Embedded Finance (escalation case)

Self-test passes 6/6 against golden outputs.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-23 19:08:31 +02:00
Benjamin Admin
e8ec50e0fc feat(control-pipeline): 24 demo test cases for applicability engine
YAML-based test package with 4 categories (6 each):
- Standard sector cases (Telko, SaaS, Energie, Automotive, Health, Law)
- Scope-beats-sector (Bank+Battery, KI-Recruiting, White-Label, Payments)
- False friends (Stripe!=PSD2, Hotline!=TKG, Repo-signals!=regulation)
- Escalation (IoT-SIM, FinTech unclear, Treuhand, KI-Diagnose)

Enforces 5 acceptance rules: no false certainty, scope>sector,
repo signals insufficient, standard first, 40%+ negative tests.

Scoring framework: must_include + must_not_include + reasoning + escalation.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-23 17:42:38 +02:00
Benjamin Admin
1f8667c7da feat(control-pipeline): replace similarity-only dedup with LLM-verified dedup in pipeline
Stage 4 (Harmonization) now uses two-tier approach:
- Score >= 0.92: auto-duplicate (embedding only, fast)
- Score 0.85-0.92: LLM verification via local qwen3.5 (think=false, ~3s)
- Score < 0.85: not a duplicate

This eliminates ~44% false positives from pure embedding similarity.
LLM_DEDUP_ENABLED env var controls the feature (default: true).

Also adds 10 applicability use case tests (bank+TAN, webshop+Stripe,
SaaS startup, energy provider, health app, automotive, law firm, etc.)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-23 16:57:37 +02:00
Benjamin Admin
441d5740bd feat: Applicability Engine + API-Filter + DB-Sync + Cleanup
Some checks failed
CI / go-lint (push) Has been skipped
CI / python-lint (push) Has been skipped
CI / nodejs-lint (push) Has been skipped
CI / test-go-consent (push) Successful in 35s
CI / test-python-voice (push) Successful in 33s
CI / test-bqas (push) Successful in 37s
CI / Deploy (push) Failing after 2s
- Applicability Engine (deterministisch, kein LLM): filtert Controls
  nach Branche, Unternehmensgroesse, Scope-Signalen
- API-Filter auf GET /controls, /controls-count, /controls-meta
- POST /controls/applicable Endpoint fuer Company-Profile-Matching
- 35 Unit-Tests fuer Engine
- Port-8098-Konflikt mit Nginx gefixt (nur expose, kein Host-Port)
- CLAUDE.md: control-pipeline Dokumentation ergaenzt
- 6 internationale Gesetze geloescht (ES/FR/HU/NL/SE/CZ — nur DACH)
- DB-Backup-Import-Script (import_backup.py)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-09 21:58:17 +02:00
Benjamin Admin
e3ab428b91 feat: control-pipeline Service aus Compliance-Repo migriert
Control-Pipeline (Pass 0a/0b, BatchDedup, Generator) als eigenstaendiger
Service in Core, damit Compliance-Repo unabhaengig refakturiert werden kann.
Schreibt weiterhin ins compliance-Schema der shared PostgreSQL.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-09 14:40:47 +02:00