AWS AI Agent Directives

AWS is broad enough that a security review either takes a specialist or never happens, and these directives make an AI agent the practical substitute. Each AWS directive gives the agent a narrow, read-only assignment backed by real aws cli calls: audit IAM policies for least privilege, investigate a cost anomaly through Cost Explorer, review security groups for open ports, enforce tag compliance, check S3 buckets for public access and encryption, find CloudTrail coverage gaps, and run an EKS best-practices review. The prompts are written for the agents people already use, so cursor rules aws and windsurf aws rules drop straight into a rules file and run against whatever account the agent is pointed at. The design principle is honest auditing: the agent lists resources, applies the documented commands, and returns a structured verdict table with the exact remediation to request, rather than asserting a clean bill of health. For anyone capturing cursor rules aws iam, aws security, or ai agent devops directives for multi-cloud work, this category is built to be run on a schedule and to hold up under review. Keep the set together for a monthly pass that returns evidence rather than opinions every time it runs. Grouped this way they turn a routine schedule into a posture review that keeps producing evidence long after the first run, and that stays legible to whoever inherits the account later.