market-research

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs agents to fetch and scrape open public sources (e.g., "WebSearch", "Firecrawl/Defuddle", "site:reddit.com", G2, Crunchbase, pricing pages) in the required agent workflows (consumer-landscape-agent, competitor-agent, trends-agent) and to read/interpret user-generated community content as evidence that directly drives gap/opportunity identification and follow-on actions, which exposes the agent to untrusted third-party content that could enable indirect prompt injection.