triaging-live-sql-activity

Pass

Audited by Gen Agent Trust Hub on Jul 22, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect prompt injection surface identified in the triage workflow.
  • Ingestion points: The skill ingests untrusted data from the database via 'SHOW CLUSTER STATEMENTS' and 'SHOW CLUSTER SESSIONS' as described in 'SKILL.md' and 'references/sql-queries.md'.
  • Boundary markers: The instructions lack explicit delimiters or 'ignore embedded instructions' warnings for the data returned from database monitoring views.
  • Capability inventory: The skill provides functionality for high-privilege administrative actions, specifically 'CANCEL QUERY' and 'CANCEL SESSION', enabling the agent to modify the state of the database cluster.
  • Sanitization: There are no documented steps for the agent to sanitize or validate query strings retrieved from the database before analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 22, 2026, 03:58 PM
Security Audit — agent-trust-hub — triaging-live-sql-activity