The Agent Skills Directory

[SAFE]: The skill provides a framework for generating bash wizards to assist humans with manual configuration steps (e.g., provisioning infrastructure or setting up API keys).
[SAFE]: The provided template.sh includes security-conscious helpers for secret management:
ask_secret: Uses read -rs to hide sensitive input from the terminal.
write_env: Creates .env files with umask 077 (mode 0600) to ensure secrets are not world-readable.
set_secret: Securely transmits credentials to GitHub Actions using the official gh CLI tool via stdin.
[SAFE]: The instructions explicitly direct the AI to use ask_secret for sensitive values and discourage running the generated scripts automatically, maintaining a human-in-the-loop requirement for browser interactions and credential handling.

wizard