The demo needs no cloud credentials: --help, then find-bills on the sample card export in docs/examples/. The sample charges are made up.
Find what you are paying for and not using, catch the cloud and developer subscriptions you forgot about, and test how your systems fail. Everything runs on your machine with your own credentials. There is no hosted backend.
Quickstart
# 1. Install: grab a prebuilt binary from the Releases page (no Go needed),
# or build it yourself with Go 1.26.10+:
go install github.com/AndrewGrayYouNeeK/Cloudripper/cmd/cloudripper@latest
# 2. Look at fictional sample output. No credentials and no network.
# The resources, costs, and savings figures are made up.
cloudripper demo
# 3. Point it at an account. AWS uses the normal SDK credential chain.
export AWS_PROFILE=my-readonly-profile
export AWS_REGION=us-east-1
# 4. Scan, then look at spend and savings ideas
cloudripper rip # inventory what is running
cloudripper cost # estimated vs. actual trailing 30-day spend
cloudripper optimize # recommendations only; nothing changes without --apply
# 5. Hunt for bills you forgot about (no cloud credentials needed)
cloudripper find-bills --sample
cloudripper find-bills --csv card-export.csv
cost and rip use AWS Cost Explorer for actual spend. Enable it first in the Billing console; data can take about 24 hours to appear. For GCP, set GCP_PROJECT; for Azure, set AZURE_SUBSCRIPTION_ID; for Cloudflare, set CLOUDFLARE_ACCOUNT_ID and CLOUDFLARE_API_TOKEN. See Configuration.
Features
- Waste scan.
rip inventories compute, disks, addresses, load balancers, databases, storage, functions, and Kubernetes across AWS, GCP, and Azure, plus Cloudflare Workers, R2, KV, D1, and Pages.
- Real spend.
cost reads trailing 30-day spend from Cost Explorer, BigQuery billing export, Azure Cost Management, or Cloudflare billing history when those APIs are available, and reports estimates and actuals separately.
- Savings recommendations.
optimize suggests deleting unattached resources, rightsizing, and scheduling. Changes need --apply plus a confirmation, honor --dry-run, and are written to an audit log.
- Forgotten bills.
find-bills reads billing mail (IMAP or Gmail, read-only) and a card CSV for AWS, Google Cloud, Azure, DigitalOcean, Linode/Akamai, Vercel, Netlify, Heroku, Cloudflare, Supabase, Firebase, MongoDB Atlas, and OpenAI, and flags the ones you have not set up.
- Scheduling, scaling, and cost alerts.
schedule, scale, and watch stop idle compute on a schedule, scale groups from CPU, and alert Slack or email on spend spikes.
- Bounded chaos tests.
chaos runs Chaos Mesh or Gremlin experiments inside a blast-radius limit, and refuses production-looking targets without --allow-prod.
- Output you can script.
--output table|json|csv|html on every command that prints rows. HTML is one self-contained file: no external assets, values are escaped, unpriced costs are n/a, and missing actuals are unavailable rather than $0.
Free and paid
The CLI is free and MIT licensed.
| Price | Status |
|---|
| CLI | Free | Available now (this repo) |
| Pro | $29/month | Coming soon. Join the waitlist |
| Team / Business | From $199/month | Get in touch |
More at cloudripper.youneek.dev.
How it works
One CLI for cloud waste, real spend, and safe chaos tests, across AWS, GCP, Azure, and Cloudflare.
It inventories resources, reads actual trailing spend when billing APIs are available, applies cost actions you confirm, scales instance groups from CPU, watches for cost spikes, and posts chaos experiments inside a blast-radius limit.
Pricing heuristics fill in when a billing API is missing or denies the call. Estimates and actuals are reported separately because they measure different things: an estimate is a monthly price guess for one resource, and an actual is account spend over the last 30 days.
Commands
cloudripper demo
cloudripper rip
cloudripper cost
cloudripper optimize
cloudripper optimize --apply
cloudripper scale
cloudripper scale --policy policy.json --apply
cloudripper watch --once
cloudripper schedule
cloudripper chaos --target my-app --kind PodChaos --duration 30s
cloudripper rip --all-regions
cloudripper find-bills --csv card-export.csv
--output table (the default), json, csv, or html works on every command that prints rows. CLOUDRIPPER_OUTPUT sets the same value when the flag is omitted. CSV cells that start with =, +, -, or @ are prefixed so spreadsheets do not treat them as formulas. html writes one document with inline CSS and no scripts, images, or network requests. Every value is escaped. An unpriced resource is n/a. Actual spend that the billing APIs did not return is unavailable, not $0.
Table output uses ANSI color only when stdout is a terminal. NO_COLOR (any non-empty value) and --no-color turn it off. json, csv, and html are never colored.
--dry-run previews mutations. optimize, schedule, and scale also require --apply before they will change anything, and --dry-run still wins if both are set. Live changes ask for one confirmation on a terminal, or accept --yes when stdin is not a terminal.
여기까지가 README 앞부분입니다. 나머지는 원문에서 읽을 수 있습니다. README 전문 보기 ↗
