Our team reviews cloud infrastructure for startups and small engineering teams. We keep finding the same categories of problems: resources that cost money without doing useful work, compute sized for traffic that never arrived, weak cost attribution, and bills that teams cannot connect to utilization.

Each environment has its own constraints. The investigation still follows a repeatable path. We collect inventory, billing data, and metrics. Then we compare cost with usage, inspect the evidence, and decide which changes make sense.

We built CloudOpt to make that analysis available to more teams. Today we are releasing its first alpha.

What CloudOpt Does Today

CloudOpt is a local, read-only, open-source cloud optimization analyzer for AWS and GCP. You run it from the command line against the scope you choose.

The current workflow has four steps:

  1. Collect cloud inventory, billing data, and utilization metrics.
  2. Store a snapshot in a local workspace.
  3. Run deterministic optimization checks against that evidence.
  4. Export the results as an HTML or JSON report.

CloudOpt looks for waste, rightsizing opportunities, idle resources, and missing cost attribution. Each finding comes from a defined check rather than a hidden recommendation engine.

You control when an assessment runs and which reports leave your machine. CloudOpt uses read-only access and does not modify cloud resources. Read-only access still requires care: use least-privilege credentials, protect the local workspace, and review reports before sharing them.

Try the Alpha

Install CloudOpt on Linux, macOS, or Windows with Git Bash:

curl -fSsL https://cloudopt.kernul.io/install | bash

The command workflow is:

cloudopt init
cloudopt collect
cloudopt analyze
cloudopt report

You can also run the bundled offline demo from the CloudOpt website without cloud credentials. The demo covers the collection, analysis, and reporting path with fixture data, so you can inspect the output before granting access to an AWS account or GCP project.

This is an alpha release. Expect gaps in provider coverage, checks, output, and documentation. We want users to see those limits and report where the tool falls short.

Why We Are Building It From Our Daily Work

Kernul works as an infrastructure team for companies that do not have one in-house. Cost optimization forms part of that work, alongside architecture, operations, security, and scaling.

That work gives us concrete inputs for CloudOpt. When we encounter a recurring cost or utilization problem, we can study the conditions that produced it and decide whether it belongs in a deterministic check. A useful check needs clear evidence, understandable logic, and a remediation path an engineer can evaluate.

We do not want to turn one unusual customer situation into a universal rule. Cloud environments carry business constraints that a scanner cannot infer from metrics alone. CloudOpt should show the evidence and help an engineer investigate. The engineer still decides whether a change is safe.

Development will continue alongside our infrastructure work. New patterns, edge cases, and reporting needs will shape the checks we add. We will treat planned work as planned work until the code and documentation ship.

Open Source Makes the Checks Better

Our team sees only a fraction of the ways companies use AWS and GCP. Public development lets platform engineers, DevOps teams, FinOps practitioners, and consultants test CloudOpt against different account structures and workloads.

We need feedback on questions such as:

  • Does a finding match the resource’s real role?
  • Does the report include enough evidence to validate the recommendation?
  • Which false positives or missed problems make the output less useful?
  • Where do setup instructions, error messages, or permissions need work?

Testing and issue reports count as contributions. So do report validation, documentation fixes, code, and helping another user reproduce a problem. The project uses the GNU GPL v3 license, and the source is available in the CloudOpt GitHub repository.

Part of a Broader Tooling Direction

CloudOpt is one focused tool in the open-source toolkit we are building for DevOps, Platform Engineering, and FinOps teams. We want each tool to solve a specific operational problem on its own.

Over time, we plan to connect these tools through a broader platform for infrastructure specialists. That platform is a direction, not a released product. CloudOpt provides one foundation through its cloud data model, collection workflow, optimization checks, and portable reports.

Publishing the alpha now gives teams a useful analyzer and gives us evidence about the workflows the wider tooling ecosystem needs.

Help Us Test the First Alpha

Install CloudOpt, run the offline demo, or test a read-only scan against AWS or GCP. Check the findings against what you know about the environment and inspect the generated report.

If a result is wrong, unclear, or missing evidence, open an issue on GitHub. Include the provider, resource type, expected result, actual result, and redacted evidence when you can. Those reports will guide the next checks and fixes.