database-lookup

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

• Third-party content exposure detected (high risk: 0.85). At runtime the skill queries multiple outsider-authored public REST APIs (e.g., NCBI/ENA/ENCODE/Eurostat/etc.) and then feeds the returned JSON fields (including free-text like titles/descriptions/notes) into the agent context as “raw JSON results,” which can contain attacker-controlled text for indirect prompt injection.