The Agent Skills Directory

[SAFE]: The skill implements robust security controls for handling API credentials. It includes logic to ensure that keys are only transmitted to authorized API origins and prevents credential leakage through redirected requests.
[SAFE]: Secure error handling practices are employed. The script uses a mechanism to clear exception contexts before re-raising errors, which prevents sensitive information (like API responses or signed URLs) from being exposed in stack traces or logs.
[SAFE]: Data validation is performed on the output from the remote provider. The script verifies image MIME types, formats, dimensions, and payload integrity using the Pillow library before saving files locally.
[SAFE]: Network operations are restricted to the OpenRouter API, and the script includes bounds on response sizes to mitigate potential denial-of-service or memory exhaustion attacks from a misbehaving provider.

nano-banana-pro