NYC

postgresql-optimization

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION] (SAFE): The skill includes SQL commands for database administration such as ALTER SYSTEM, CREATE INDEX, and VACUUM. While these commands require high database privileges and can impact performance (e.g., exclusive locks during VACUUM FULL), they are standard for the skill's intended purpose of PostgreSQL optimization and are provided as documentation/examples.
  • [DATA_EXPOSURE] (SAFE): No hardcoded credentials or sensitive system file paths were detected. Example configuration files for PgBouncer use standard non-sensitive paths (e.g., /etc/pgbouncer/userlist.txt) as templates.
  • [INDIRECT_PROMPT_INJECTION] (LOW): The skill involves analyzing database outputs such as query plans and statistics, which is a potential surface for indirect injection if database metadata is attacker-controlled.
  • Ingestion points: Database output (EXPLAIN ANALYZE results, pg_stat_statements).
  • Boundary markers: Absent.
  • Capability inventory: Database configuration (ALTER SYSTEM), extension management, and index creation.
  • Sanitization: None described in the templates.
  • Note: This is a characteristic of the use case rather than a malicious defect.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:21 PM