verify-results

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

• Third-party content exposure detected (high risk: 0.85). Outsider free text can enter the LLM context via the runtime-provided paper.tex/--input and the artifact repo files that extract_claims.py and audit_repo.py read as plain text (e.g., tables/README authored by others), which the agent then summarizes into the LLM context; this is not limited to user-authored content.