I was manually checking for idle AWS instances every month. So I built a tool.

Vu Huu Phuong — Fri, 06 Feb 2026 09:10:22 +0000

Every month I'd log into AWS console, click through each region, check CPU metrics, make a spreadsheet. Took 1-2 hours.

The problem? I was only checking EC2 instances. Volumes, snapshots, load balancers—I'd look at those maybe twice a year when the bill seemed high.

Last month I finally did a full audit. Found an EBS volume we forgot about - $120/month for 8 months. That was the breaking point.

Spent a weekend building a CLI that checks everything automatically. Ran it on our account: $800/month in waste. Ouch.

"But AWS has Cost Explorer and Trusted Advisor?"

Yeah, I know. And Azure has Cost Management. GCP has Recommender.

Here's why I still built this:

Trusted Advisor's free tier is limited. The good stuff (right-sizing recommendations) needs Business/Enterprise support. That's $100+/month minimum.

Cost Explorer shows spending, not why. You can see your EC2 bill went up, but you still have to dig through CloudWatch metrics manually to find which instances are idle.

Multi-cloud is painful. We use AWS and Azure. Switching between portals, different UIs, exporting to spreadsheets to compare... it's a mess.

I wanted something fast. Run one command. Get a list. Done.

So yeah, the cloud providers have tools. But they're either expensive, buried in consoles, or split across platforms.

What I built

A CLI that scans AWS, Azure, and GCP for waste:

npm install -g cloud-cost-cli
cloud-cost-cli scan --provider aws --region us-east-1

It finds:

Idle EC2/VM instances (low CPU for 7 days)
Unattached EBS volumes and disks
Snapshots older than 90 days
Oversized RDS databases
Unused load balancers
And 15+ other things

(Screenshot from a different account - I didn't capture my original scan, but the format looks like this)

What it found in our account

Real examples from our first scan:

$240/month - EBS snapshots created every 12 hours, kept forever. We forgot to set retention.

$180/month - Staging environment running 24/7. Nobody told me we only use it 9-5 weekdays.

$120/month - Volume attached to an instance we terminated 8 months ago. AWS doesn't auto-delete volumes when you kill instances. Learned that the hard way.

$85/month - EC2 instance at 8% CPU. Used to be a CI runner, migrated to GitHub Actions, forgot to turn it off.

$175/month - RDS instance in a region we don't even use anymore. From a failed migration attempt last year.

Total: $800/month = $9,600/year

These aren't huge numbers individually. But they add up. And we're a small team. I don't want to know what enterprises are leaking.

The dashboard (just added)

CLI works great for automation, but I wanted something visual.

Built a web dashboard:

cloud-cost-cli dashboard

Opens at localhost:9090. You can:

Store credentials (encrypted locally)
Run scans with a button
See trends over time
Filter by confidence level

It doesn't send data anywhere. Everything stays on your machine. I'm not interested in your cloud bills.

Multi-cloud is the real win

Honestly, if you only use AWS and have Business support, Trusted Advisor might be enough for you.

Where this actually helps: when you use multiple clouds.

We use AWS (main infra) and Azure (some client projects). I can run:

cloud-cost-cli scan --provider aws --all-regions
cloud-cost-cli scan --provider azure --all-regions

And get one unified list. Same format. Same confidence scoring. Actual apples-to-apples comparison.

That's what I couldn't get from the native tools.

How it works (brief version)

Uses the official SDKs:

AWS SDK (CloudWatch for metrics, EC2 for resources)
Azure SDK (Monitor, Compute, Storage)
GCP SDK (Compute, Monitoring)

For right-sizing recommendations, it pulls 7 days of CPU/memory/network/disk metrics. If everything's consistently low (< 10%), it suggests downsizing.

Confidence levels:

HIGH - All metrics low, safe to act
MEDIUM - Multiple metrics low, worth reviewing
LOW - CPU-only or mixed signals, manual check needed

All read-only. It won't touch your infrastructure.

Try it

GitHub: https://github.com/vuhp/cloud-cost-cli

npm install -g cloud-cost-cli
cloud-cost-cli scan --provider aws --region us-east-1

Or launch the dashboard:

cloud-cost-cli dashboard

It's free and open source. MIT license.

What's next

Still pretty early. Things I want to add:

Export reports to PDF/Excel (for finance team)
Historical comparison (track progress over time)
Alert when waste spikes
Maybe scheduled scans

But honestly, it already does the main thing: finds waste fast.

If you're tired of cloud bills creeping up, give it a shot. Worst case, you find out you're actually optimized (nice problem to have). Best case, you find money sitting there doing nothing.

Let me know if you find anything interesting. Or if it breaks. Probably still has bugs.

Forem: Vu Huu Phuong