aws-iam-policy-analysis
Warn
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes Python scripts that programmatically interact with AWS IAM and S3 services to perform account modifications. Specifically, the enumeration script in 'references/s3-attack-techniques.md' executes logic to create new IAM roles ('iam.create_role'), new IAM users ('iam.create_user'), and generates associated access keys ('iam.create_access_key').
- [DATA_EXFILTRATION]: The skill's methodology and scripts utilize 'webhook.site' to receive AWS service callbacks and metadata. The SNS bypass script in 'references/aws-resource-policy-exploits.md' automatically confirms subscriptions by sending requests to this external, non-whitelisted service, facilitating the capture of SNS message details.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a high vulnerability surface for indirect prompt injection. It ingests untrusted IAM policies, Lambda source code, and CloudFormation templates as data. It lacks boundary markers or sanitization logic to prevent malicious instructions embedded in these audited files from influencing the agent's behavior, particularly given its capabilities for network access and IAM manipulation.
- [PRIVILEGE_ESCALATION]: The skill provides specific methodology and automated logic for identifying and exploiting IAM privilege escalation paths, such as 'sts:AssumeRole' and 'iam:PassRole', which are used to transition between identities and increase permissions within the target AWS environment.
Audit Metadata