The Agent Skills Directory

[SAFE]: No malicious behavior, direct prompt injection, or obfuscation techniques were detected. The skill is focused on legitimate Amazon SNS administration.
[COMMAND_EXECUTION]: The skill utilizes a tool named SNSTool to execute AWS SNS commands such as CreateTopic, ListSubscriptions, and DeleteEndpoint. These operations are appropriate for the skill's stated purpose of managing cloud notifications.
[DATA_EXFILTRATION]: The Publish command enables sending messages to external endpoints (SMS, email, etc.). While this involves network transmission, it is the fundamental purpose of the SNS service and is documented here for administrative use.
[CREDENTIALS_UNSAFE]: Commands like CreatePlatformApplication include parameters for platform credentials (e.g., PlatformCredential). The skill provides the structure for these calls using placeholders and does not include any hardcoded sensitive information.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes untrusted data (e.g., message bodies in the Publish command or topic names in CreateTopic) which are then used in cloud administrative actions. No boundary markers or sanitization steps are specified within the skill instructions to mitigate this risk.

awsclaw-sns