Skip to main content
ChangeGuard is change intelligence for Kubernetes. It continuously tracks every deployment, config, and GitOps change across your fleet, correlates each failing workload to the change that caused it, explains the root cause with cited evidence, and helps your team decide what’s safe to ship — before, during, and after production changes.
Who these docs are for. Platform/Kubernetes operators (install & operate) and developers / on-call engineers (day-to-day use). If you’re evaluating ChangeGuard for your organization, start at changeguard.ai.

​
What you get today

Change Intelligence

Every failing workload linked to the change — and the diff — that shipped it. GA

Incident investigation

Root cause with cited evidence, a proposed fix, and how to verify it. GA

CSC Score & pre-flight

A deterministic deploy-readiness score and a SHIP / HOLD / BLOCK verdict — advisory. GA

Autonomous Remediation

Policy-governed detect → fix → verify → rollback, at the autonomy level you choose. GA · opt-in
Maturity labels. We mark capabilities GA (generally available), Early Access (design partners / on request), or Roadmap. The Engineering Advisor, Engineering Opinion / Experience / Reasoning, and Multiplayer AI are Early Access — powerful, but not yet self-serve. We never document an emerging capability as if it ships today. See Product concepts.

​
How it works, in one line

A lightweight, read-only collector runs in your cluster (installed by the ChangeGuard operator) and pushes cluster state and your changes over HTTPS to the ChangeGuard control plane, which does the correlation, scoring, and analysis. Your source code, secret values, and workload data stay in your cluster. See Architecture & data flow.

​
Time to value

ChangeGuard is advisory by default (autonomy level Advise). It changes none of your workloads until you explicitly opt into remediation and grant the RBAC.