address-review

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

• Third-party content exposure detected (high risk: 0.70). SKILL.md describes triaging “findings” that may arrive as a PR/PR comment or session context, which are outsider-authored free-form text, and the workflow then “presents each finding” to the agent for dispositioning—implying that such text is ingested into the LLM context at runtime.