aws-fis-experiment-prepare
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches experiment templates from official AWS documentation pages (e.g., docs.aws.amazon.com/en_us/fis/latest/userguide/az-availability-scenario.html) to serve as the authoritative source for multi-action template structures.
- [COMMAND_EXECUTION]: The skill executes local shell scripts included in the package to manage infrastructure. Specifically, it uses 'scripts/deploy-with-retry.sh' for CloudFormation stack deployment, 'scripts/precheck-cfn-permissions.sh' to verify IAM permissions via policy simulation, and 'scripts/rename-output-dir.sh' for directory management.
- [DYNAMIC_EXECUTION]: The agent dynamically generates infrastructure-as-code (cfn-template.yaml) that includes IAM roles and inline Python Lambda functions (for EKS RBAC management and resource tagging) by assembling provided reference skeletons with discovered resource metadata.
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection attack surface by processing data from external documentation and possessing deployment capabilities.
- Ingestion points: Reads AWS documentation pages via the agent's documentation tool to extract JSON templates (SKILL.md).
- Boundary markers: The instructions do not specify boundary markers or instructions to disregard embedded commands for the ingested documentation content.
- Capability inventory: The skill has the ability to write local files, execute shell commands through scripts, and deploy CloudFormation stacks creating IAM roles and Lambda functions.
- Sanitization: There is no explicit sanitization or validation of the extracted documentation content before it is used to generate the CloudFormation template.
Audit Metadata