The Agent Skills Directory

[SAFE]: The skill follows a well-structured, multi-phase approach that prioritizes user intent and verification. Key security-positive attributes include:
Secret Management: Explicitly instructs the user to add local setting files (.local.md) to .gitignore to prevent credential exposure.
Human-in-the-Loop: Requires explicit user confirmation after discovery and planning phases before generating any code or executing commands.
Least Privilege Guidance: Recommends using the minimal necessary tools and provides a security-first principle checklist (e.g., enforcing HTTPS and forbidding hardcoded credentials).
Safe Command Execution: Uses standard development utilities like mkdir and git init for project scaffolding, with no evidence of dangerous remote code execution or exfiltration techniques.

create-plugin