Qwen 3.5 uses all tokens for thinking, leaving response empty.
Using /no_think prefix to get direct JSON output.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Qwen 3.5 with latest Ollama returns structured thinking in separate
'thinking' field, leaving 'response' empty. Now checks both fields.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replaces scroll+filter approach with proper semantic search:
1. Embed query via bp-core-embedding-service (bge-m3, 1024 dim)
2. Vector search in Qdrant (bp_compliance_datenschutz + bp_compliance_gesetze)
3. Sort by cosine similarity score
4. No API key needed — local Qdrant on Mac Mini
Falls back gracefully: SDK first, then semantic Qdrant, then empty.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Go SDK points to external Qdrant (qdrant-dev.breakpilot.ai) with expired API key.
Fallback: search directly in local Qdrant (bp-core-qdrant:6333) which has
all collections: bp_compliance_datenschutz, bp_compliance_gesetze, atomic_controls_dedup.
Search strategy:
1. Try Go SDK RAG endpoint (preferred, has embedding-based search)
2. Fallback: Qdrant scroll with text-based regulation filter
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New module: rag_document_checker.py
- Searches RAG (Qdrant) for controls relevant to document type
- Filters by regulation (DSGVO Art.13, TDDDG §25, BGB §355 etc.)
- LLM (Qwen 3.5:35b) verifies each control against document text
- Returns fulfilled/missing with evidence text + severity
- Supports: DSI, Cookie, Impressum, Widerruf, AGB, DSFA, AVV, Loeschkonzept
Integration in doc-check endpoint:
- Regex checklist runs first (fast, deterministic)
- RAG checks run after (semantic, catches what regex misses)
- Both results combined in single response
LLM prompt returns JSON: {fulfilled, evidence, issue, severity}
Think-tags stripped, JSON extracted from response.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Only Cookie and Widerruf sections are checked as separate documents.
Social Media, DSFA, Betroffenenrechte, Dienste von Drittanbietern are
part of the parent DSI and no longer generate false findings.
Added PLAN-rag-document-check.md for Phase 2:
- RAG-based checks with document-type-specific Controls
- DSFA checklist (Art. 35 + Landes-Listen)
- AVV checklist (Art. 28)
- Reference detection (sub-doc → parent doc)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When a single URL contains multiple document sections (e.g. IHK DSI page
with Cookies, Social Media, Dienste von Drittanbietern), the system now:
1. Extracts full page text (main document check as before)
2. Splits text at heading boundaries (short uppercase lines)
3. Classifies each section: Cookie→cookie checklist, Social Media→DSI etc.
4. Runs type-specific checklist per section
5. Returns all results: main doc + sub-sections
Section type detection via SECTION_TYPE_MAP patterns:
- 'Cookie*' → §25 TDDDG checklist
- 'Dienste von Drittanbietern' → DSI checklist
- 'Social Media' → DSI checklist (Art. 26 joint controllership)
- 'Widerrufsrecht' → §355 BGB checklist
- 'Impressum' → §5 TMG checklist
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New "Dokumenten-Pruefung" tab in Compliance Agent:
- User adds multiple URLs with document type (DSI, AGB, Impressum, Cookie, Widerruf)
- Each document loaded via Playwright, accordions expanded, text extracted
- Checked against type-specific legal checklist
- Optional: Cookie banner check via checkbox
Checklisten-UX (solves "100% looks like nothing was checked"):
- All checks shown per document: green checkmark + matched text excerpt
- Red X for missing fields with legal reference
- Builds user trust: "9 Punkte geprueft, alle bestanden"
- Expandable per document with completeness bar
New checklists:
- Impressum: §5 TMG (6 fields: name, address, contact, register, VAT, representative)
- Cookie-Richtlinie: §25 TDDDG (5 fields: types, purposes, retention, third-party, opt-out)
Backend:
- POST /agent/doc-check — async with polling (same pattern as /scan)
- DocCheckResult includes checks[] with passed/failed + matched_text
- dsi_document_checker returns all_checks in SCORE finding
- Email report shows per-document checklist
Files: agent_doc_check_routes.py (280 LOC), DocCheckTab.tsx (248 LOC),
ChecklistView.tsx (130 LOC), dsi_document_checker.py (+70 LOC)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Service detection:
- Only search script tags + src/href attributes for service patterns
- Prevents false positives from DSE text mentioning services
(e.g. IHK DSE describes etracker, 'google analytics' in text)
- Technical patterns (with regex chars) still checked in full HTML
Short documents:
- Documents with < 200 words flagged as 'Kurzhinweis' instead of
'MANGELHAFT' — too short for Art. 13 completeness check
- Prevents 96-word navigation pages from showing 8 missing fields
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Two fixes:
1. consent-tester: full_text truncation raised from 10,000 to 50,000 chars
(IHK Internetangebot has ~50K chars, Beschwerderecht was after 10K cutoff)
2. Backend: dse_text now combines Playwright HTML + ALL DSI discovery texts
for mandatory content checking. Previously only used first 8K chars from
one source, missing Verantwortlicher/DSB that were in DSI documents.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Root cause: When all 9 Art. 13 checks passed (100%), no SCORE finding
was created (line: 'if pct < 100'). The backend then defaulted to
completeness=0 because it looked for the SCORE finding to extract the %.
Fix: Always generate SCORE finding, even at 100%. Added 'OK' severity
for fully compliant documents.
This was the cause of 8 documents showing '0% MANGELHAFT' despite
containing all required information.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
DSI Dedup (consent-tester):
- Only H1/H2 headings count as documents (not H3/H4 sub-sections)
- Sub-sections (Cookies, Betroffenenrechte, Social Media) are part of
parent document's full text, not separate documents
- Reduces IHK result from 30 to ~11 real documents
Backend (agent_scan_routes):
- ScanFinding gets doc_title field linking each finding to its document
- doc_title set when creating DSI findings for document attribution
Frontend (ScanResult.tsx):
- 3 sections: Services table, Document cards, General findings
- Documents: expandable cards with completeness bar (green/yellow/red)
- Findings grouped under their parent document
- Each card shows: title, word count, findings count, % completeness
- Findings without doc_title go to "Allgemeine Findings" section
Email Summary (agent_scan_helpers):
- Findings listed under their parent document
- General findings in separate section
- No more flat mixed list
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Fundamental fix: scans now run asynchronously with progress polling.
Backend:
- POST /scan starts background task, returns scan_id immediately
- GET /scan/{scan_id} returns status + progress + result when done
- 7 progress steps shown: Website scan, DSI discovery, DSE analysis,
SOLL/IST comparison, corrections, report, email
- In-memory job store (dict with scan_id → status/result)
- No timeout limits on scan duration
Frontend:
- POST starts scan, receives scan_id
- Polls GET every 5 seconds (max 120 attempts = 10 min)
- Shows live progress message during scan
- Displays result when completed, error when failed
Proxy:
- POST timeout reduced to 30s (just starts the job)
- GET timeout 10s (just status check)
- No more 504/connection-dropped errors
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Bug 1: max_pages was hardcoded to 15 in backend call — raised to 50
Bug 2: DSI documents checked against text_preview (500 chars) — now uses
full_text (10,000 chars) for Art. 13 mandatory field checks
Bug 3: DSE text not found when Playwright misses DSE page — now falls
back to DSI Discovery full_text as second source
Bug 4: Backend timeout 120s too short for 50 pages — raised to 300s
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
These files existed on the feature branch but were never cherry-picked
to main, causing ModuleNotFoundError on import:
- dse_parser.py — parses DSE HTML into structured sections
- dse_matcher.py — matches detected services against DSE sections
- mandatory_content_checker.py — checks Art. 13 DSGVO mandatory fields
- legal_basis_validator.py — validates legal basis (lit. a-f)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
NameError: name 're' is not defined at line 146 — the import was
accidentally removed when extracting helper functions to agent_scan_helpers.py.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- CookieBannerOverlay: shows vendors per category with expandable tables
(Verarbeiter, Cookies, Dauer, Land) for full transparency
- Demo vendors: 4 necessary, 3 statistics, 3 marketing, 3 functional
- cookie_table_generator.py: renders {{COOKIE_TABLE}} Markdown tables
from vendor configs (DB) or service registry (fallback)
- SERVICE_COOKIES: 16 known vendor-to-cookie mappings with provider + country
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Email now lists all scanned URLs with checkmark/cross status.
Frontend shows collapsible "X Seiten gescannt — Details anzeigen".
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
SDK LLM chat returns empty content due to Qwen think-mode. Direct Ollama
/api/generate call with stream:false gets the full response including
think tags which we strip.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Backend: mode field in request, adapts summary tone and email subject
- Pre-launch: "Implementieren Sie X vor Veroeffentlichung"
- Post-launch: "ACHTUNG: Maengel sind oeffentlich sichtbar, sofortige Nachbesserung"
- Frontend: Mode toggle (internes Dokument vs. Live-Website)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Scan public website for cancellation button, imprint, privacy link, cookie consent
- Generate follow-up questions when checks can't be verified without login
- User answers "no" → finding with legal basis is added to results
- Frontend: FollowUpQuestions component with Ja/Nein buttons
- Sidebar: "Compliance Agent" entry added under KI-Compliance
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
SessionLocal: 5x verwendet fuer DB-Sessions ausserhalb Depends()
HTTPException: verwendet in Framework-Validation
text: 55x verwendet fuer raw SQL queries
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Python: add missing 'import enum' to compliance/db/models.py shim.
TypeScript: remove duplicate export of useVendorCompliance from
vendor-compliance/context.tsx (already exported from ./hooks).
Docs: add mandatory pre-push checklist (lint + test + build) to
AGENTS.python.md and AGENTS.go.md. [guardrail-change]
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- loc-budget CI job: remove if/else PR-only guard; now runs scripts/check-loc.sh
(no || true) on every push and PR, scanning the full repo
- sbom-scan: remove || true from grype command — high+ CVEs now block PRs
- scripts/check-loc.sh: add test_*.py / */test_*.py and *.html exclusions so
Python test files and Jinja/HTML templates are not counted against the budget
- .claude/rules/loc-exceptions.txt: grandfather 40 remaining oversized files
into the exceptions list (one-off scripts, docs copies, platform SDKs,
and Phase 1 backend-compliance refactor backlog)
- ai-compliance-sdk/.golangci.yml: add strict golangci-lint config (errcheck,
govet, staticcheck, gosec, gocyclo, gocritic, revive, goimports)
- delete stray routes.py.backup (2512 LOC)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Benjamin Admin
sharang