monitoring-database-health

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill is configured with restricted bash access for database clients, specifically psql, mysql, and mongosh. This is necessary for its core function of monitoring database health and performing remediation steps.
  • [PROMPT_INJECTION]: This skill interacts with untrusted data retrieved from external database systems (e.g., query results, logs, or metrics). This creates a surface for indirect prompt injection where malicious instructions stored in a database could influence the agent's behavior. The provided instructions lack explicit boundary markers or sanitization logic for processing this external data.
  • Ingestion points: Reads database metrics and status via psql, mysql, and mongosh tools referenced in SKILL.md.
  • Boundary markers: None identified in the instruction set or logic.
  • Capability inventory: Has Write, Edit, and Bash access to execute database commands as defined in SKILL.md.
  • Sanitization: No specific sanitization or validation of database query output is mentioned.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 11:34 AM