Files
breakpilot-compliance/zeroclaw/docs/batch-test-results-2026-05-08.md
T
Benjamin Admin 1b5c6bd340
Build + Deploy / build-ai-sdk (push) Failing after 33s
Build + Deploy / build-developer-portal (push) Successful in 7s
Build + Deploy / build-tts (push) Successful in 7s
Build + Deploy / build-document-crawler (push) Successful in 7s
Build + Deploy / build-dsms-gateway (push) Successful in 8s
Build + Deploy / build-admin-compliance (push) Successful in 1m51s
Build + Deploy / build-backend-compliance (push) Successful in 8s
CI / loc-budget (push) Failing after 18s
CI / secret-scan (push) Has been skipped
CI / go-lint (push) Has been skipped
CI / python-lint (push) Has been skipped
CI / nodejs-lint (push) Has been skipped
CI / validate-canonical-controls (push) Successful in 19s
Build + Deploy / build-dsms-node (push) Successful in 8s
CI / branch-name (push) Has been skipped
Build + Deploy / trigger-orca (push) Has been skipped
CI / guardrail-integrity (push) Has been skipped
CI / nodejs-build (push) Successful in 3m8s
CI / dep-audit (push) Has been skipped
CI / sbom-scan (push) Has been skipped
CI / test-go (push) Failing after 46s
CI / test-python-backend (push) Successful in 41s
CI / test-python-document-crawler (push) Successful in 32s
CI / test-python-dsms-gateway (push) Successful in 24s
docs: Batch test results for 9 websites + EUIPO analysis
Tested BMW, Stadt Koeln, BfDI, Sparkasse, Caritas, TUEV Sued,
Spiegel, ETO Gruppe, EUIPO. Key findings:

- Stadt Koeln + ETO Gruppe best (95% correctness)
- BMW, Sparkasse, Spiegel genuinely deficient (verified)
- EUIPO uses EU Regulation 2018/1725, not GDPR — needs separate checklist
- ~0-2 false positives per website after LLM verification

7 regex fixes emerged from batch testing (soft hyphens, word
insertions, numbered headings, German section names, etc.)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-05-08 00:41:28 +02:00

2.0 KiB

Batch-Test Ergebnisse (2026-05-08)

9 Websites getestet

# Website Typ L1 L2 Vollst. Korr. Woerter Bewertung
1 Stadt Koeln Kommune 9/9 21/22 100% 95% 5910 Vorbildlich
2 Caritas Nonprofit 9/9 19/22 100% 86% 9447 Gut
3 ETO Gruppe Mittelstand 9/9 21/22 100% 95% 7312 Vorbildlich
4 BfDI Bundesbehoerde 9/9 16/22 100% 73% 2014 OK (kurz)
5 TUEV Sued Prueforg. 8/9 15/21 89% 71% 9467 Luecken
6 IHK Konstanz Kammer 9/9 18/22 100% 82% 6353 Gut
7 BMW Konzern 8/9 10/21 89% 48% 7207 Mangelhaft
8 Sparkasse Finanz 7/9 10/20 78% 50% 12183 Mangelhaft
9 Spiegel Medien 6/9 10/13 67% 77% 13698 Mangelhaft

Sonderfaelle

  • EUIPO (EU-Behoerde): 6/9 L1, 5/13 L2 — unterliegt Verordnung 2018/1725, nicht DSGVO. Separate Checkliste noetig.
  • dm, Zalando, HWK: Text-Extraktion scheitert (JS-heavy SPAs, Consent-Wall blockiert)

Verifizierte True Positives

BMW, Sparkasse und Spiegel haben tatsaechlich lueckenhafte DSEs — verifiziert gegen Originaltexte:

  • BMW: Keine E-Mail, kein Art. 77 Beschwerderecht, keine Art.-Referenzen fuer Rechte
  • Sparkasse: Kein DSB, kein Art. 77
  • Spiegel: Kein DSB, kein Art. 77, keine Betroffenenrechte

False-Positive-Rate

Ueber alle 9 Websites: ~0-2 FP pro Website nach LLM-Verifikation. Hauptursache verbleibender FP: Ungewoehnliche Formulierungen die weder Regex noch LLM erkennen.

Regex-Fixes die aus dem Batch-Test entstanden

  1. Soft-Hyphen Stripping (\xad) — etogruppe
  2. "Recht auf [Adjektiv] Auskunft" — Wort-Einschub
  3. "nach Fortfall" neben "nach Ablauf" — Loeschkonzept
  4. DSB-Kontakt ueber Zeilenumbrueche — [\s\S]{0,300}
  5. Nummerierte Headings ("5. Soziale Medien") — isdigit()
  6. Section-Splitter nur bei klassifizierten Headings
  7. "Soziale Medien/Netzwerke" als Social-Media-Heading