The Agent Skills Directory

[SAFE]:
No malicious patterns or security vulnerabilities were identified.
The skill design incorporates 'Anti-patterns' that prevent unsafe tool configurations, such as the use of unrestricted API calls.
[COMMAND_EXECUTION]:
Executes standard GitHub CLI (gh) commands for repository management.
Implements a strict permission model by leveraging allowed-tools to limit the agent's scope to specific subcommands.
[PROMPT_INJECTION]:
Mandates user confirmation for all 'Write' operations, preventing the agent from performing potentially harmful actions autonomously.
Instructions are focused on task execution and do not attempt to bypass system constraints.

github