Mutova

Velos · Infrastructure Remediation Engine

Turn cloud optimization findings into reviewable infrastructure patches.

Velos takes a structured cloud cost recommendation and an Infrastructure-as-Code repository, identifies the relevant resource definition, proposes the smallest required code change, validates the patch, and opens a draft pull request for engineering review.

Cloud finding→Repository analysis→Infrastructure patch→Validation→Draft pull request

How Velos works

1

Finding received

Import a recommendation from a FinOps tool, cloud scanner, or structured JSON input.

↓
2

Repository understood

Velos searches Terraform or other Infrastructure-as-Code files and identifies the affected resource, module, variables, and dependencies.

↓
3

Patch generated

AI proposes a minimal, scoped infrastructure change based on the remediation rule and repository context.

↓
4

Safety validated

Deterministic pre-checks verify the live infrastructure state, while formatting, validation, tests, policy checks, and infrastructure plans inspect the proposed patch.

↓
5

Draft PR opened

Velos creates a reviewable pull request containing the patch, assumptions, pre-check evidence, expected savings, and validation results.

Example mission

Example mission

Remove an unused EBS volume

  • Confirm the volume is still unattached
  • Verify it has remained unused for the required period
  • Check account, region, tags, backup policy, and ownership
  • Locate the Terraform definition
  • Remove the resource with a minimal patch
  • Run terraform fmt, terraform validate, and terraform plan
  • Open a draft pull request for human approval

AI proposes the patch. Velos enforces the safety boundary.

AI is used for repository understanding and patch generation. Authorization, infrastructure pre-checks, scope restrictions, validation, and pull-request creation remain deterministic.