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>
Qwen 3.5:35b needs ~30-60s per call. Multi-call scan was timing out. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- website_scanner.py: multi-page crawl, 20+ service patterns (tracking, CDN, chatbots, payment, fonts, captcha, video), AI text detection - dse_service_extractor.py: LLM extracts services from privacy policy text - agent_scan_routes.py: POST /agent/scan — combines scan + DSE comparison, generates findings (undocumented, outdated, third-country transfer), auto-corrections via Qwen in pre-launch mode Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
