scrape-explore-site

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

• Third-party content exposure detected (high risk: 0.85). Outsider free text can enter the LLM context indirectly because download.py fetches arbitrary runtime web pages (start/list/detail URLs) and writes their readable HTML to rendered.html, which extract_links.py then parses and emits link anchor text (link["text"]) into groups.txt for the agent to read/classify.