market-research

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

• Third-party content exposure detected (high risk: 0.90). The skill's orchestrator and multiple agents (notably competitor-agent and consumer-landscape-agent) explicitly require live WebSearch / scraping tools (Firecrawl/Defuddle) and sources like pricing pages, G2/Crunchbase, Reddit and other social/forums, and instruct the agent to read and use that user-generated public content as evidence that drives gap identification and opportunity ranking, so it clearly ingests untrusted third-party content that can change behavior.