The Agent Skills Directory

[SAFE]: The skill provides security-hardened IAM templates that align with AWS best practices. These include the use of condition keys like sts:ExternalId to prevent confused deputy attacks and aws:PrincipalOrgID for simplified cross-account access within an organization.
[COMMAND_EXECUTION]: The skill includes reference CLI commands for common IAM auditing tasks, such as generating credential reports and listing access keys. These actions are standard administrative procedures consistent with the skill's purpose.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to review and troubleshoot IAM configurations, creating a surface for processing untrusted user data.
Ingestion points: User-supplied IAM policy JSON and AWS configuration data provided for review.
Boundary markers: Absent; there are no specific instructions for the agent to treat policy content as non-executable data.
Capability inventory: The agent utilizes Bash with aws CLI tools as specified in the allowed-tools configuration.
Sanitization: No specific sanitization or schema validation for ingested policies is mentioned.

iam