de-cost-optimization

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

• Third-party content exposure detected (medium risk: 0.65). The skill’s runtime workflow includes querying public/outsider-authored free text from cloud query history/system tables (e.g., system.runtime.query_history / system.query_log selecting query/query_snippet), where the query field contains arbitrary SQL text originally submitted by users outside the operating user and is then ingested into the agent’s LLM context.