validate

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs the agent to "Research the problem" by finding evidence from public, user-generated sources such as Reddit/forum threads, Twitter/X, competitor reviews and community links (see the "Tell AI" prompts and "Where to look for evidence" section), requiring the agent to fetch and interpret untrusted third-party content that can influence validation decisions.