I shipped a small thing: a browser-based AWS waste scanner
Just me, solo dev. Spent the last few months building CloudSaver. Drop your AWS cost CSV, it runs locally and gives you a safe-delete list: what's waste, the order to delete it in, and what to check before you touch anything. No upload, no auto-delete. It's open source and free. If you want me to sanity-check your plan, I offer a $19 review. Still early. Would love feedback from anyone who's cleaned up AWS bills before. try it: https://getcloudsaver.com/?src=ih
The strongest part is the safety boundary: analyze locally, produce a deletion plan, but never auto-delete. That makes the tool useful for finding AWS waste without asking users to hand over credentials or blindly trust automation.
Thanks aryan. that was the whole point really. small teams get the same safety net the big shops have, without handing over creds or auto-deleting stuff.
sure thing. support@getcloudsaver.com is the best one. on real usage feedback so far: mostly small stuff. grouping at 200+ resources is the open problem, not safety. would love to hear what you'd want if you ran it on your own accounts
Thanks! I’ve just sent it over.
Looking forward to hearing your thoughts whenever you have a chance.
got it, will check beryxa when i get a chance. heads down on shipping for now.
Running everything locally without uploading anything is a strong trust signal for a tool that touches AWS billing data. Have you tested it against larger accounts (hundreds of resources) yet, or is it still tuned for smaller setups?
Good question. tested on a few hundred resources, no issues yet. rules get noisy at that scale though, still working on grouping. what's your setup like?
Makes sense, grouping is the hard part once volume goes up — flat rule lists don't scale past a certain point. My setup's smaller scale (backend + game logic services), but I went through the same "rules get noisy" phase with anything that fires per-resource instead of per-pattern. Curious if you're grouping by tag, by resource type, or something smarter — that's usually where these tools live or die for me
Honestly right now it's mostly resource type. tag-based is on the list but not done. i'm with you, that's where it lives or dies. / what's your game stack? curious what kind of resources you end up with at the messy end.
Node/Express + Postgres (Prisma) on the backend, Angular on the front, a handful of cron jobs driving game rounds. The messy end for us is mostly stale/orphaned state from those crons — content banks running dry, rounds that should've auto-resolved and didn't, that kind of thing. Less "unused EC2 instance" and more "background job that silently stopped doing its job." Different flavor of waste, same instinct to go hunt for it
this is the part that gets overlooked. the delete-order list is what makes cleanup safe. without it everyone just leaves it alone and the bill grows quietly. the solo dev at $40/mo is the whole pitch in one number. forgotten infra is the most expensive kind because nobody's mad enough to go hunting for it. glad it landed. if you want to poke at the actual scanner, it's here: https://getcloudsaver.com/?src=ih (free, runs in your browser, no creds). curious whether your own waste ends up more "orphaned EIPs" or "jobs that silently stopped" once you run it.
To add context: the scanner's main job is catching the boring waste nobody notices. orphaned elastic IPs, 90-day-old snapshots, idle load balancers. one solo dev i know was paying ~$40/mo for stuff he'd forgotten existed. the delete-order list is what makes it safe to actually clean up.