KvantumCI Release

Overworld

Out now v1.3.0 Released June 2026

The map just got bigger. Overworld extends verification to AWS and the ML supply chain, sharpens the rule engine, and scales to organizations with thousands of repositories.

Explore the expansion

Introducing Update 1.3

Your world just got bigger. Overworld brings verification to AWS and Bedrock, delivers model security for DevOps teams, expands the rule engine, and scales cleanly to organizations with thousands of repositories.

What’s in the expansion

Five upgrades, one Overworld

Everything that ships in v1.3.0—follow the rail through each chapter below.

01 AWS Support Scan AWS agents, guardrails, and Bedrock. 02 Model Security MLSecOps rules and MLflow best practices. 03 Rule Engine New ML/Nvidia rules, faster reviews. 04 Verification UI Clearer reads, new run statuses. Finale Scale & Performance Thousands of repos, faster dashboards.

01Cloud coverage

AWS Support

Overworld lands on AWS. Bring cloud-native AI workloads inside your verification perimeter—scan AWS agents, detect guardrails, and review AWS Bedrock usage.

  • AWS agent scanning across pipelines
  • Guardrails detection for AWS workloads
  • AWS Bedrock usage review
AWS agent scanning, guardrails, and Bedrock review
MLSecOps rules and model security for DevOps teams

02MLSecOps

Model Security

Security for the ML supply chain, built for DevOps teams. New MLSecOps rules and MLflow best practices catch risky model workflows before they ship.

  • MLSecOps rules and MLflow best-practice checks
  • Model security for DevOps teams
  • New technology detections across the ML stack

03Detection engine

Rule Engine

The detection engine gets sharper and faster. New rules cover MLOps, Nvidia tooling, MLflow, and model scanning, with CPU-based optimization for quicker code reviews.

  • New rules for MLOps, Nvidia tooling, MLflow, and model scanning
  • CPU-based optimization for faster code reviews
  • Improved detection markers and SCA/SAST tooling fixes
  • Better GitLab pathname handling and error management
Expanded rule engine for MLOps and Nvidia tooling
Improved verification UI with new run statuses

04Experience

Verification UI

A clearer read on every run. UI improvements sharpen readability across the product, and verification runs gain new statuses so you always know where a check stands.

  • UI refinements for better readability
  • New statuses for verification runs
  • Clearer signal across the verification views

FinaleScale

Scale & Performance

Overworld is built for the biggest estates. Repo pagination now loads organizations with thousands of repositories, and the dashboard is faster and more precise.

  • Repo pagination for orgs with thousands of repositories
  • Dashboard performance optimization for faster loading
  • More precise dashboard data

Overworld is live in v1.3.0. Map your estate and verify your next run today.

Explore the Overworld

Verify your AWS and ML workloads, run the expanded rule engine, and scale across thousands of repositories—on your next run. No credit card required.

Patch notes · v1.3.0

Overworld — full patch notes

Everything that shipped in the Overworld release.

New features

  • AWS support: AWS agent scanning, guardrails detection, and AWS Bedrock review.
  • MLSecOps: New MLSecOps rules and MLflow best practices, plus new technology detections.
  • Model security: Model security brought to DevOps teams.
  • UI improvements: Better readability across the product and new statuses for verification runs.

Rule engine

  • New rules: Coverage for MLOps, Nvidia tooling, MLflow, and model scanning.
  • CPU-based optimization: Faster code reviews.
  • Detection improvements: Fixed problems with some SCA and SAST tooling and improved markers in our detection capabilities.
  • GitLab: Better pathname handling and improved error management.

Fixes

  • Repo pagination: Fixed pagination so we can load more repos for organizations with thousands of repositories.
  • Dashboard performance: Loading optimization and more precise data.
See the v1.3.0 entry in the changelog →