aws-secrets-manager
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions and scripts involve ingesting data from an external source (AWS Secrets Manager), creating a surface where malicious instructions embedded in secrets could influence agent behavior.
- Ingestion points: Data is retrieved from AWS Secrets Manager using the AWS CLI and Python
boto3SDK as demonstrated in the rotation and integration examples withinSKILL.md. - Boundary markers: The provided code examples and instructions do not include specific delimiters or guidelines to distinguish the retrieved secret content from system-level instructions.
- Capability inventory: The skill has access to the shell (via AWS CLI) and network capabilities (via
boto3andpsycopg2libraries) to interact with AWS services and databases. - Sanitization: There is no explicit logic shown for sanitizing or validating the retrieved secret values before they are used in downstream operations such as connection string construction or database queries.
Audit Metadata