postgres-pro

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • Prompt Injection (SAFE): The instruction set in SKILL.md focuses on role-definition and task constraints without attempting to bypass safety protocols or extract system prompts.
  • Data Exposure & Exfiltration (SAFE): No hardcoded secrets, API keys, or sensitive local file paths were found. The references to remote servers in the postgres_fdw documentation use clearly marked placeholders ('remote_host', 'remote_password').
  • Remote Code Execution (SAFE): There are no patterns involving the download and execution of scripts from external URLs. The skill describes standard PostgreSQL extension installation via SQL (CREATE EXTENSION) and routine DBA tool usage (pg_repack).
  • Obfuscation (SAFE): The content is entirely human-readable markdown and SQL. No Base64, zero-width characters, or homoglyphs were detected.
  • Indirect Prompt Injection (LOW): The skill possesses a surface for indirect injection because it processes user-provided SQL queries for optimization. However, it lacks dangerous capabilities like automatic command execution on the host system, mitigating the risk level to LOW.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:42 PM