What is built, and what is not
Most roadmaps are a wish list with the failures deleted. This one comes from the codebase.
How to read this. Everything under “in the console today” is running code you can be shown in a demo. Everything under “next” and “later” is not built yet, and we have not attached dates because we would only be guessing. The build log records what actually landed and when.
In the console today14
You can open these screens right now. Nothing here is a promise.
CoverOps Marketplace
Free, open-source DevOps kits at marketplace.coverops.dev. Six AWS kits were applied, re-planned with no changes and destroyed on a real AWS account before we called them tested; the date is on each kit.
Azure without a stored secret
Your Azure app registration trusts CoverOps by workload identity federation. A client secret still works as a fallback.
Live audits on AWS, Azure, GCP, DigitalOcean, Oracle Cloud and Cloudflare
Connect an account with the read-only role CoverOps generates (Terraform, CloudFormation or CLI). Inventory, utilisation and cost come from the provider's own APIs, and every figure says whether it came from billing data or list prices.
Self-serve signup and the audit report
Sign up and connect one account for a free scan: your total waste and the top three findings. The audit lists every finding, priced per month in ₹ or $, with the evidence and the fix, reviewed by an engineer within 3 working days.
The console: cloud accounts, findings, approvals, audits and billing
Cloud inventory and cost, findings priced in rupees, approvals with rollback, the audit with its report, GST invoices and credit notes, a data export, and staff visits only with a reason. Kubernetes, GPU and agent screens run on demo data only.
Approved fixes with rollback
On AWS: stop, rightsize and delete-volume proposals on resources you tag coverops:managed=true. Each is re-checked against the live resource at approval and keeps a snapshot to roll back from.
Alerting across 26 metrics
One rule builder over infrastructure, cost and security metrics, sent in-app, by browser push, email, Slack, webhook and PagerDuty. GPU and agent metrics come from the demo estate only.
Off-hours, storage tiering, NAT and commitments
New findings on AWS: dev and test servers running all night, gp2 disks that should be gp3, large S3 buckets with no lifecycle rule, idle NAT gateways, cross-zone transfer, and what AWS itself recommends you commit to. Each comes with its command, a dry run and an undo; your team applies it.
Savings measured on your bill
Each fix marked done is checked by the next scan, then measured on your cloud bill 7 and 30 days later. Cost Explorer reports by service and region, so a measured saving is never credited above the fix's own estimate.
AWS Organizations and one-click setup
Launch the read-only role in AWS with one click, your External ID filled in. Connect the management account once and add its member accounts from a list, with one StackSet for all of them.
Agent plan, in the console
A page in the console, not software in your account. It works from what CoverOps has already read through your read-only access and lays out what to do now, this week and this month, with the exact command for each fix. It never changes anything itself.
Cloudflare, live
Read-only scan of Cloudflare zones, load balancers, Workers and R2 buckets through an API token with read permissions only. Finds a paid plan still billing on a zone that is paused, moved or never activated, and load balancers that serve nothing. Advice-only. Proven on our own account first (17 Nov 2026).
DigitalOcean and Oracle Cloud, live
Read-only inventory, billed cost and CPU utilisation through a scoped, expiring token on DigitalOcean and a signed API key on Oracle Cloud. Being proven on the first real accounts. Fixes on these two clouds are advice-only for now.
Alerts by email, Slack, webhook and PagerDuty
Alerts now send for real, as well as in-app and browser push. Every webhook destination is checked before anything is sent to it, and a failed delivery tells you why.
Next1
Being built now.
Azure and Google Cloud kits, tested
Azure dev-VM auto-shutdown and Google Cloud cost guardrails are published as untested; they become tested only after a sandbox run on a real subscription and project.
Later3
Real intent, no date. Better undated than an invented quarter.
In-account agent (planned)
A small agent you run inside your own cloud account, so findings are worked out there and only the results leave it. Not built yet; nothing is installed in your account today.
WhatsApp, Microsoft Teams and SMS
Approvals and alerts where Indian teams already talk. Not being built yet; these channels are not connected, and the console says so.
SOC 2 programme
We are not certified and will not imply otherwise. Today the platform keeps a named approval on every change and an audit log you can export as JSON.
Deliberately never1
Not a gap. A boundary, and the reason the rest is safe to adopt.
Hosting your workloads
This is a boundary, not a gap. CoverOps reads and changes your cloud; it never hosts it. The moment we run your workloads, leaving us stops being easy.
What would change our order
This is our current read, not a commitment. What moves an item up is hearing it is the reason a team cannot adopt us. If your blocker is a provider we do not cover live, or a certification we do not hold, say so. Knowing which gap costs us the most conversations is how we decide what to build next.
Tell us what is blocking you, or read the platform overview for what the four domains do today.