postgres

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/pgdiag, the required runtime workflow ingests outsider-authored free text via the user-provided --plan-for argument and then runs EXPLAIN (FORMAT JSON) {query} through psql, meaning attacker text can directly reach the LLM context/processing path at runtime.