Amazon Q Developer

If your infrastructure lives on AWS, Amazon Q Developer is the assistant that actually knows AWS — not from generic training, but grounded in the live service catalog and, when connected, your own account.

What sets it apart

The grounding is the differentiator. Ask a general model “why is my ECS task stuck in PENDING?” and you get plausible guesses. Ask Q Developer the same with your account connected and it can reason about your task definition, the actual subnet/security-group wiring, and the specific console error — then point at the real cause. It also generates CloudFormation, CDK, Terraform, and AWS CLI that tends to be current and idiomatic, because keeping up with AWS’s release pace is exactly its job.

The agentic features are real: /dev implements a scoped change across files, /transform upgrades Java/.NET versions, /review does a code review pass, and /test scaffolds tests. There’s also a security scan that catches the over-broad IAM policy or public S3 bucket before it ships.

Where it shines for DevOps

• AWS IaC — first-draft CloudFormation/CDK/Terraform you then review against a Terraform plan-review discipline.
• CLI archaeology — “what’s the AWS CLI to list every unattached EBS volume with its cost?” lands fast.
• EKS and AWS-native ops — troubleshooting cluster IAM, IRSA, and load-balancer controller issues.
• Security review — flagging risky IAM and exposed resources at authoring time.

Where to be careful

• Treat /dev output like any AI change: read the diff, run the plan, never auto-apply. The human-in-the-loop rule applies — Q proposes, you approve, and it never holds the credentials that mutate prod.
• Off-AWS, reach for Claude (reasoning/postmortems) or Cursor (multi-file repo edits) instead.
• Double-check service details on the newest releases; even AWS’s own assistant occasionally lags its own changelog.

How to get the most out of it

• Connect it to your account (with read-scoped permissions) so answers are grounded, not generic.
• Use the IDE plugin for completion and the chat for “explain this AWS error.”
• Keep destructive actions behind a human and a plan — use Q to draft the change, your pipeline to apply it.

Pricing notes

The free tier covers a lot of individual daily use. Pro at $19/user/month unlocks higher limits and the fuller agent feature set — reasonable if AWS is your primary platform and the IaC/code-gen time savings are real.