aws-solution-architect
Pass
Audited by Gen Agent Trust Hub on Apr 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes several Python scripts (
architecture_designer.py,serverless_stack.py,cost_optimizer.py) designed to be executed by the agent. These scripts perform functional logic such as cost estimation and template generation without involving dangerous system calls or external network requests. - [PROMPT_INJECTION]: The skill processes user-supplied application requirements (e.g., application type, expected scale, budget) to generate architectural recommendations and infrastructure-as-code (IaC) templates.
- Ingestion points: Requirements are ingested via JSON in
architecture_designer.pyandserverless_stack.py. - Capability inventory: The generated templates include IAM roles, DynamoDB tables, and Lambda functions.
- Sanitization: Basic sanitization is performed on application names (e.g., lowercasing and replacing spaces), though most user-provided strings are interpolated directly into the generated output templates. This represents a standard surface for indirect prompt injection where a user could influence the generated IaC, but the risk to the agent itself is minimal.
- [DATA_EXPOSURE]: The skill explicitly instructs against hardcoding credentials and provides examples of using AWS Secrets Manager for secure credential retrieval in the
references/best_practices.mdfile.
Audit Metadata