The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from external SQS messages which could influence agent behavior.
Ingestion points: ReceiveMessage command in SKILL.md reads message bodies from remote queues.
Capability inventory: The skill possesses destructive and sensitive capabilities such as DeleteQueue, PurgeQueue, SendMessage, and AddPermission in SKILL.md.
Boundary markers: No delimiters or warnings are provided to scope the processing of received message content.
Sanitization: No validation or filtering logic is specified for the ingested message data.
[COMMAND_EXECUTION]: The skill provides tools for performing administrative and destructive operations on AWS SQS resources.
Evidence: Includes commands like DeleteQueue to remove infrastructure, PurgeQueue to delete all message data, and AddPermission to modify resource-based access policies.

awsflow-sqs