What problem does it solve? AI agents frequently produce incorrect IAM policies and roles — wrong action names, missing confused deputy protections, unsafe condition operators, and hallucinated API behaviors. This Skill provides verified corrections for these edge cases plus structured workflows for role management and baseline policy generation. ## Core Features & Use Cases - Verified IAM Edge Cases: Corrections for STS session limits, Organizations quirks, SAML/MFA specifics, CloudTrail logging behavior, and policy evaluation subtleties like ForAnyValue/ForAllValues Null checks. - Role Management Workflows: Create and maintain service roles and execution roles with proper trust policies, aws:SourceArn/aws:SourceAccount confused deputy conditions, and scoped permissions. - Policy Generation: Generate baseline IAM policies from application source code or a Terraform plan JSON using iam-policy-autopilot, with a Service Authorization Reference fallback for unsupported languages. - Use Case: A developer provides Python source code using boto3 and asks for a least-privilege IAM policy — the Skill routes to the Autopilot tool with correct flags instead of hand-writing a potentially hallucinated policy. ## Quick Start Ask the assistant to generate an IAM policy from your application source code or Terraform plan JSON, or to create a service role for an AWS resource like a Glue crawler or Lambda function.