What this prompt does
This prompt casts the model as a senior AWS architect and FinOps engineer and asks it to audit an account for cost savings with concrete, prioritised findings instead of generic advice. You feed it four context values — [account_profile], [monthly_spend], [top_services], and [constraint] — and it works through EC2 right-sizing, a waste sweep, commitment coverage, storage tiering, and network costs, then ranks everything by savings-to-effort.
The structure works because it separates the cost problem into the categories where AWS bills actually leak, then forces a prioritised table at the end. [monthly_spend] and [top_services] tell the model where the money concentrates, so it spends its attention on EC2 and RDS rather than rounding errors. The [constraint] field is what keeps the recommendations usable — if you state no downtime windows or a multi-AZ production requirement, the model won't suggest a fix that violates them. [account_profile] grounds the audit in what the account actually runs, so the right-sizing and tiering advice reflects your real environments rather than a generic textbook account. Ranking by savings-to-effort at the end is what turns a long list of possibilities into a plan you can act on this week.
When to use it
- Your AWS bill creeps up every month and nobody has audited it in a while.
- You're heading into a Savings Plan or Reserved Instance renewal and want quick wins first.
- You suspect over-provisioned EC2 instances but lack a structured way to confirm it.
- You want to find unattached EBS volumes, idle RDS, and stale snapshots in one pass.
- You need a prioritised list to take to stakeholders, ranked by effort and risk.
- You're reviewing S3 storage classes and lifecycle rules for tiering opportunities.
Example output
Expect a prioritised checklist with a findings table at its core: each row a finding, its estimated monthly saving, the effort and risk to implement, and a one-line remediation step. Findings are grouped by the six audit areas — EC2 right-sizing, waste sweep, commitment coverage, storage tiering, and network costs — and ranked so the highest savings-to-effort items sit at the top.
Pro tips
- Fill
[top_services]from your actual Cost Explorer breakdown so the audit targets the real spenders. - Be honest in
[constraint]— a fix that needs downtime is worthless if you stated production must stay multi-AZ. - Validate NAT gateway and egress findings against real CloudWatch numbers before committing; estimated savings there are the easiest to overstate.
- Keep
[monthly_spend]approximate but current, so the percentage savings the model implies are grounded. - Run it again after implementing the top tier to surface the next layer of savings.
- Treat commitment recommendations as a prompt to model your own usage; don't buy a Savings Plan straight from an estimate.
- Cross-check the waste sweep manually — an EBS volume that looks unattached may belong to a stopped instance someone plans to restart.
- Feed it the constraint that matters most to your stakeholders; a finance-driven audit and an engineering-driven one weight effort and risk very differently.
