The Agent Skills Directory

Prompt Injection (SAFE): No attempts to override agent behavior or bypass safety filters were detected. The documentation uses standard instructional language for technical guidance.
Data Exposure & Exfiltration (SAFE): No hardcoded credentials or sensitive file paths (like ~/.aws/credentials) are present. Database connection examples use generic placeholders such as 'username', 'hostname', and 'database_name'.
Obfuscation (SAFE): No Base64, zero-width characters, homoglyphs, or encoded commands were found in the README.md file.
Unverifiable Dependencies & Remote Code Execution (SAFE): The skill does not perform any package installations or remote script execution (e.g., curl | bash). References to external resources point to the official postgresql.org domain.
Privilege Escalation (SAFE): No use of sudo, chmod 777, or other privilege-modifying commands. Database role management is discussed as a standard DBA task, not as an exploit.
Persistence Mechanisms (SAFE): No modifications to shell profiles, cron jobs, or system services are suggested or implemented.
Indirect Prompt Injection (LOW): As a documentation skill for database operations, it defines a surface area for processing database-related tasks.
Ingestion points: User prompts requesting SQL generation or database information.
Boundary markers: Absent in this README, but typically handled by the host agent environment.
Capability inventory: File reads/writes (via \copy and -f flags) and database modification capabilities are described as intended features of psql.
Sanitization: Not applicable to documentation, but the skill advises on best practices for database security.

postgresql-psql