audit-engine

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

• Third-party content exposure detected (high risk: 0.70). The required workflow reads repository-authored governance/docs content (e.g., CLAUDE.md/AGENTS.md, docs/repo-health/governance.md, prior reports) and feeds it into the LLM during Stage 1 “Detect (LLM-first)”, which is outsider-authored free text if the repo content wasn’t authored by the operating user.