The Agent Skills Directory

[SAFE]: The skill provides comprehensive architectural guidance for AWS based on the Well-Architected Framework, emphasizing security principles like least-privilege IAM and data encryption.
[COMMAND_EXECUTION]: The skill instructs the agent to use a local shell command (ls) to detect recommended companion skills in specific directories. This is a functional feature for skill management and does not pose a security risk.
[SAFE]: The included evaluation suite (evals.json) specifically tests the agent's ability to identify and warn against insecure practices, such as hardcoding credentials or exposing databases to the public internet.

cloud-aws