iam-review
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill contains defensive instructions that explicitly tell the agent to treat analyzed content as untrusted data and to ignore any directives (like 'ignore previous instructions') found within the IAM policies it reviews. These are security safeguards rather than malicious injections.
- [DATA_EXPOSURE]: The skill includes a 'Prompt Injection Safety Notice' that strictly prohibits the exfiltration of credentials, API keys, or secrets discovered during the review, requiring the agent to redact or generically reference them in the output.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest untrusted IAM configuration data via the
$ARGUMENTSvariable. It implements a secure evidence chain by providing clear boundary markers in the 'Injection Hardening' section and limiting agent capabilities to read-only assessment tools (Read, Grep, Glob). - [REMOTE_CODE_EXECUTION]: No mechanisms for remote code execution, dynamic code loading, or unauthorized package installations were identified in the skill instructions or metadata.
Audit Metadata