Prompt
Do I need drift detection if changes are only made through Git?
Latest observation
Nov 5, 2026
OpenAI APIWeb search: off
Not always, but Git-only changes reduce the need for drift detection — they don’t eliminate it.
If Git is your only deployment path
If every change to infrastructure/configuration:
- goes through Git,
- is reviewed,
- is applied by automation (CI/CD, GitOps),
- and no one changes resources manually,
then drift risk is much lower. In that case, drift detection may be less critical than in environments with lots of manual changes.
Why it can still be useful
Even with Git-only processes, drift can still happen because of:
- Out-of-band changes by admins or services
- Provider-side mutations (cloud services modifying resources)
- Failed or partial deployments
- Human error in what got applied vs. what’s in Git
- External systems updating state/configuration
- Default/implicit settings changing over time
When you should still have it
Drift detection is especially valuable if:
- your infrastructure is security-sensitive,
- you need auditability/compliance,
- you have multiple teams or accounts,
- resources are long-lived,
- or you rely on IaC to be the source of truth.
Practical rule
- Small, tightly controlled GitOps environment: drift detection is nice to have, not mandatory.
- Anything production, regulated, or complex: drift detection is still worth it.
Bottom line
Git-only change management lowers drift risk, but drift detection is still a good safety net.
If you want, I can help you decide based on your setup (Terraform, CloudFormation, Kubernetes, etc.).