lead

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md/loop.md read GitHub issue bodies and comments as the “coordination log” and dispatch the frozen decomposition/respawn context into reviewers/builders, so outsider-authored issue/comment free text can reach runtime LLM inputs via the tracking issue/sub-issue thread.