01Finding explanations
Sharper reasons, confidence labels, and caution notes for idle compute, unattached storage, snapshots, IPs, load balancers, and Kubernetes findings.
Roadmap
Practical improvements before platform sprawl.
Cloud Waste Scanner will stay focused on a tight operating loop: local scan, evidence review, owner handoff, and verified closure. This page is written as a product commitment, not a wishlist.
Now
Make the first scan easier to trust and easier to hand off.
Next
Deepen expensive resource coverage without changing the local trust model.
Later
Add automation only where review gates and rollback context are clear.
Shipping
Evidence and first-run quality
The scanner should be useful before a team creates a program around it.
02Export packages
PDF, CSV, and handoff manifests that help finance, engineering, and platform owners review the same evidence without chasing screenshots.
03Setup diagnostics
Clearer feedback for credential scope, proxy routing, failed scan checks, and cases where no useful data was collected.
Planned
Expensive waste coverage
Depth matters most where the monthly bill can move materially.
04Kubernetes and GPU evidence
Persistent volume drift, LoadBalancer services, node baselines, GPU reservations, and owner gaps brought into the same local review queue.
05Provider-specific checks
More precise review paths for managed databases, storage lifecycle drift, network billing artifacts, and regional resource leftovers.
06Review grouping
Better grouping by owner, account, provider, resource type, priority, and confidence so teams can split review work cleanly.
Researching
Automation with guardrails
Automation should remove repeat work after the evidence model is strong enough.
07Local API workflows
Scheduled local scans, internal reporting pipelines, and export automation without requiring a hosted collector.
08Execution plan review
Approval notes, rollback context, masked exports, and policy simulation before recurring cleanup workflows are used.
09Team operating model
Owner queues, closure tracking, reopened findings, and audit-friendly evidence history for teams that repeat the process weekly.
No hosted collector requirement for the core scan path. Credentials and raw scan output stay on the operator machine unless the operator chooses to export.
No blind deletion workflow, no vague “AI optimization” button, and no broad platform expansion before local scan and handoff quality are excellent.
Operator feedback wins when it reduces review time, lowers cleanup risk, or makes evidence easier to defend in a real team process.
Influence the queue
Send the blocked review, not a feature slogan.
The most useful feedback is concrete: the provider, resource type, review step, evidence gap, and what decision the team could not make.