product-marketing-context

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 ingest and use public website content and user-generated reviews/social posts (e.g., "Live URL or public presence — use the public site, pages, and visible messaging" and Section 9: "Pull from reviews... Reddit / Twitter / G2 / Capterra / Trustpilot"), so it clearly consumes untrusted third‑party content that can influence downstream decisions and outputs.