aws-agentic-ai
Pass
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: Technical analysis confirms that the skill is dedicated to legitimate AWS infrastructure management. No obfuscation, persistence mechanisms, or data exfiltration patterns were found.
- [EXTERNAL_DOWNLOADS]: External references are limited to official AWS documentation, AWS CLI references, and recognized developer resources like OpenTelemetry and MDN. The skill mentions standard packages such as the 'bedrock-agentcore' SDK and '@apidevtools/swagger-cli'.
- [COMMAND_EXECUTION]: Automation scripts ('deploy-template.sh' and 'validate-deployment.sh') utilize standard AWS CLI and CDK commands for resource provisioning and deployment verification. These actions are performed within the user's AWS environment.
- [PROMPT_INJECTION]: The skill identifies capability surfaces for processing external data in the Browser and Code Interpreter services and mandates mitigation strategies. Ingestion points: Web content extracted via the Browser service and third-party API responses via Gateway targets. Boundary markers: Documentation recommends rigorous input validation and the use of explicit delimiters. Capability inventory: Execution of code within isolated sandboxes and full browser automation. Sanitization: README files explicitly instruct developers to sanitize extracted data and user-provided code.
- [CREDENTIALS_UNSAFE]: No hardcoded secrets, API keys, or private keys were found. The skill emphasizes the use of encrypted credential providers and AWS Secrets Manager for secure storage.
Audit Metadata