The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several local shell scripts to perform administrative tasks, including CloudFormation deployment, permission verification, and directory management.
Evidence: scripts/deploy-with-retry.sh, scripts/precheck-cfn-permissions.sh, and scripts/rename-output-dir.sh are invoked during the workflow.
[REMOTE_CODE_EXECUTION]: The skill generates CloudFormation templates that include inline Python code for AWS Lambda functions. These functions are deployed and executed in the user's AWS environment to manage Kubernetes RBAC resources and perform resource tagging.
Evidence: references/eks-pod-action-guide.md contains an inline Python Lambda function using botocore.signers.RequestSigner to manage EKS ServiceAccounts and Roles.
Evidence: references/az-power-interruption-guide.md includes logic for a Lambda function to apply tags to EC2, RDS, and ElastiCache resources.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests data from external AWS documentation to extract experiment templates, which is then processed by the agent.
Ingestion points: SKILL.md instructs the agent to use aws___read_documentation on URLs like https://docs.aws.amazon.com/en_us/fis/latest/userguide/az-availability-scenario.html.
Boundary markers: Absent; there are no explicit delimiters or instructions to ignore embedded commands within the fetched documentation.
Capability inventory: The skill has extensive capabilities including AWS CLI execution, CloudFormation stack deployment, and IAM role creation.
Sanitization: Absent; the agent is tasked with extracting JSON templates directly from the raw documentation text.
[EXTERNAL_DOWNLOADS]: The skill fetches configuration guidelines and experiment templates from official AWS documentation domains.
Evidence: Documentation URLs for Scenario Library scenarios are listed in SKILL.md and README.md.

aws-fis-experiment-prepare