bake-the-brief

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md describes an agent workflow that reads a “brief” or “close-out” that is “pasted in, or in the repo” (and the close-out is re-read in the next session), so outsider-authored free text submitted as those artifacts is ingested by the LLM at runtime before drafting.