saas-study

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). Phase 2/3 captures and LLM-ingests outsider-authored HTML/DOM and extracted markdown from URLs the user supplies (e.g., evaluate_script: () => document.documentElement.outerHTML and then extract_html.py/extract_html_buffer.py), plus it then has Claude read those extracted files to write GAMEPLAN.md (Phase 2.5) and BRIEF.md (Phase 4).