postgres-pro

Pass

Audited by Gen Agent Trust Hub on Jun 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill documents several CLI commands for database management, including service control ('systemctl stop postgresql'), promotion ('pg_ctl promote'), and high-availability tooling ('pg_autoctl'). These operations are consistent with the skill's primary purpose of PostgreSQL administration.
  • [EXTERNAL_DOWNLOADS]: Recommends the installation of database extensions and management tools such as 'pg-auto-failover' and 'pg_repack' via system package managers like 'apt-get'. These are recognized community tools for PostgreSQL.
  • [PROMPT_INJECTION]: The skill analyzes external data from database system views like 'pg_stat_activity' and 'pg_stat_statements'. This represents an indirect prompt injection surface where query strings or application names could contain instructions. The skill mitigates this by instructing the agent to use prepared statements and standard verification steps.
  • Ingestion points: 'pg_stat_statements', 'pg_stat_activity' (SKILL.md, references/maintenance.md).
  • Boundary markers: None.
  • Capability inventory: Subprocess calls for 'systemctl', 'pg_ctl', and 'apt-get' (references/replication.md).
  • Sanitization: Explicitly instructs to use prepared statements to prevent SQL injection (SKILL.md).
  • [DATA_EXFILTRATION]: Documents the use of 'postgres_fdw' (Foreign Data Wrapper) to connect to remote-host databases. This allows for legitimate cross-database querying and is documented with standard placeholder credentials.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 14, 2026, 10:32 AM
Security Audit — agent-trust-hub — postgres-pro