deep-research

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s required runtime workflow performs breadth scans and then deep reads by scraping and extracting full page content from outsider-authored web results (candidate URLs) using tools like search/search_web followed by scrape/read_url_content and even search_code/get_file_contents from GitHub repos/issues/PRs.