The Agent Skills Directory

[COMMAND_EXECUTION]: Several scripts (secure_setup.sh, secure_setup.py, add_client.py, cleanup_client.py, and get_admin_conn.sh) execute external CLI tools such as doctl, gh, and psql to manage infrastructure.
These executions are the primary function of the skill and are performed using structured command lists (e.g., subprocess.run(['doctl', ...])) rather than raw shell strings.
Input variables used as SQL identifiers (like schema names or usernames) are validated against a strict regular expression (^[A-Za-z_][A-Za-z0-9_]*$) to prevent injection attacks.
[CREDENTIALS_UNSAFE]: Static detection flagged potential hardcoded connection strings in scripts/secure_setup.py and reference/path-b-schema-isolation.md.
Detailed review confirms these are non-sensitive templates and documentation examples using placeholders like {username}:{password} or CHANGE_ME_SECURE_PASSWORD.
The skill does not contain any actual hardcoded secrets.
[DATA_EXFILTRATION]: The skill handles sensitive database credentials during setup processes.
Passwords are generated using cryptographically secure methods (secrets module in Python, openssl in shell).
The 'Hands-Free' setup scripts are designed to transmit generated passwords directly to GitHub Secrets using the gh CLI, ensuring credentials are never printed to the console or stored in local logs.
[SAFE]: The skill is authored by an official DigitalOcean entity and exclusively interacts with verified services and tools. The use of bindable variables for managed databases represents a high-security pattern that eliminates the need for manual password handling by the user.

postgres