Live demo · mock data
A read-only scan of your AWS, GCP & Azure accounts for orphaned, idle and over-provisioned resources — ranked, priced in £/mo, with the exact fix.
Reclaimable / mo
£225
scanning…
Findings
7
Write actions
0
§ 01 / One-Command Setup
Don't even read the docs — paste this into Claude Code, Cursor, or any AI terminal. It pulls the open-source agent, scans every project read-only, and pushes your waste straight to the console.
Set up FeckBills to find wasted spend in my cloud.
Pull the open-source, read-only agent, scan every project I can
see, then push the findings to my console:
docker run --rm -v ~/.config/gcloud:/root/.config/gcloud:ro \
ghcr.io/feckbills/feckbills-agent \
scan --all-projects --push --token $FECKBILLS_API_KEY
Use my gcloud application-default credentials (read-only, never write).
When it's done, open feckbills.com/console and show me the 5 biggest leaks.
Swap $FECKBILLS_API_KEY for your own — a free key is one click away in the console.
§ 02 / The Console
Every scan becomes living history: savings reclaimed over time, waste by category, continuous watch across all your accounts.
§ System / Overview
Reclaimable / mo
£0
Findings
0
Saved to date
£0
Savings reclaimed
last 8 months
Latest scan
liveGKE requests ≫ usage (9% P95 CPU)
Idle EC2 (m5.xlarge, peak 3% CPU)
Unattached 1 TiB gp3, 41d
RDS idle, 0 connections 14d
Unassociated Elastic IP, billed idle
§ 03 / Namespace X-ray
For GKE we map every namespace's real activity — used CPU, network, and request traffic — into a heatmap. Spot the ones reserving compute but doing nothing, and claw the spend back. Flagged, never auto-killed: you decide.
Namespace activity · acme-prod
used CPU over 14d
2 idle namespaces · £312/mo reserved compute doing nothing.
§ 04 / Detector Array
We stand on each cloud's own metrics & recommendation APIs, then add the "lost resource" detectors they miss — and rank everything by £ impact.
AWS · GCP · Azure
Pods reserving 4 vCPU and burning 0.15 at P95 — the wedge, priced as reclaimable capacity.
Disks nobody deleted — incl. leaked pvc-* volumes — still on the bill 30/60/90 days on.
External IPs you're billed for precisely because they sit unattached.
Snapshots whose source disk is gone, or quietly ageing past 90 days.
Persistent disks still billing on TERMINATED Compute Engine instances.
Running VMs with near-zero CPU over the window (gke-* nodes excluded).
§ 05 / Global Sweep
Every region, every cloud — GCP, AWS and Azure, each in its own colour. Bigger blob, bigger monthly leak.
§ 06 / Connect Sequence
Grant a least-privilege role or run the open-source agent in your own account. No write scopes, ever.
We pull the cloud's own metrics & recommendations, then run our detectors for the waste they miss.
A ranked, plain-English action list in £/mo — what to change, the saving, and the risk.
No secrets, no write scopes, no raw resource data leaves your account — only findings. The agent is open source so you can audit exactly what it reads before you connect it.
Run the free CLI against your own account, or connect read-only and let the console watch the waste for you.