The Agent Skills Directory

[Prompt Injection] (SAFE): No malicious instructions, override patterns, or bypass markers were detected in SKILL.md or metadata. \n- [Indirect Prompt Injection] (LOW): The skill defines ingestion points for untrusted data via the table_name and schema parameters in SKILL.md. While boundary markers are not explicitly defined in the templates, sanitization is robustly enforced through strict regex patterns (e.g., ^[a-z][a-z0-9_]*$). The skill lacks autonomous execution capabilities, functioning primarily as a template generator. \n- [Dynamic Execution] (LOW): The utility script scripts/validate.py and provided documentation in PATTERNS.md correctly utilize yaml.safe_load() for configuration parsing, effectively mitigating unsafe deserialization risks.

postgresql-fundamentals