anysite-market-research

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly instructs the agent to fetch and analyze untrusted public sources—e.g., search_reddit_posts, search_twitter_posts, search_linkedin_posts, parse_webpage, duckduckgo_search, and sec_document—so user-generated and web content is ingested and can directly influence the agent's decisions and actions.