The Agent Skills Directory

[SAFE]: The skill follows secure credential management by sourcing API keys from environment variables or local configuration files, and it includes logic to mask these keys in error messages to prevent accidental exposure.
[SAFE]: Network activity is confined to the official Stability AI API domain. The implementation includes a custom HTTP handler that explicitly blocks redirects for authenticated requests, protecting against potential Man-in-the-Middle or redirection attacks targeting credentials.
[SAFE]: Output operations are restricted to a specific data directory with safe file permissions (0700), preventing unauthorized access to generated images and metadata by other users on the same system.
[SAFE]: Dependency management is restricted to well-known and trusted Python libraries (Pillow), and the code does not perform any dynamic execution, obfuscation, or persistence-related activities.

stability-ai