aws-iam-best-practices
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection due to its processing of untrusted IAM policy documents and environment security reports.\n
- Ingestion points: Data is ingested via shell commands like
aws iam get-policy-versionandaws iam get-credential-reportinSKILL.md, as well as through theboto3library in the providediam-hardening.pyscript.\n - Boundary markers: The skill lacks explicit boundary markers or instructions for the agent to ignore potentially malicious instructions embedded within the analyzed policy documents.\n
- Capability inventory: The skill includes functions capable of modifying the environment, specifically the ability to deactivate or delete IAM access keys.\n
- Sanitization: The skill utilizes structured parsing via
jq,awk, andboto3for format validation, but it does not perform sanitization or verification of the logical content within the policies themselves to prevent adversarial manipulation of the model's reasoning.
Audit Metadata