engineer

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md requires reading a Linear/GitHub ticket’s free-form body/title/comments at runtime (outsider-authored issue text) and then using it to drive the agent’s actions, which implies that this prose is fed into the LLM context via the tracker “get_issue/list_comments” path.