The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill documents commands to retrieve active session tokens (gh auth token) and set repository or environment secrets (gh secret set). While standard administrative features, these represent capabilities for handling sensitive credentials.
[COMMAND_EXECUTION]: Includes documentation for potentially destructive commands such as deleting repositories (gh repo delete), removing secrets (gh secret delete), and closing issues or PRs.
[EXTERNAL_DOWNLOADS]: Provides instructions for installing GitHub CLI extensions (gh extension install), which involves downloading and executing third-party code within the user's CLI environment.
[DATA_EXFILTRATION]: Documents patterns for programmatic access to repository data and metadata through the REST and GraphQL APIs (gh api), as well as downloading workflow artifacts (gh run download).
[REMOTE_CODE_EXECUTION]: References the action-tmate GitHub Action, which provides an interactive SSH session for debugging Actions runners. This is a common but high-privilege debugging tool.
[SAFE]: The hardcoded database connection string (postgres://postgres:test@postgres:5432/testdb) identified in the documentation is a placeholder for a local Docker service container defined within the same example and does not pose a security risk.

github