postgresql-cli

Pass

Audited by Gen Agent Trust Hub on Jul 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill acts as an educational and reference resource for the psql tool, describing its object inspection, formatting, and data import/export capabilities without including malicious instructions.
  • [COMMAND_EXECUTION]: Documents psql's shell-interaction features, such as the shell escape command (\!), piping output to shell commands (\g | command), and executing programs via \copy, while providing clear security caveats regarding command injection risks.
  • [CREDENTIALS_UNSAFE]: Provides connection guidance that explicitly warns against insecure practices like hardcoding passwords in connection strings or using the PGPASSWORD environment variable, recommending the secure ~/.pgpass method instead.
  • [PROMPT_INJECTION]: Discusses the risks of SQL injection during variable interpolation and instructs the agent to use safer alternatives such as single-quoted string substitution (:'varname') or truly parameterized queries using \bind.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 2, 2026, 02:58 PM
Security Audit — agent-trust-hub — postgresql-cli