iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
A technical report describes CVE-2026-90970 as a critical prompt-template sandbox escape that can lead to command execution in a self-hosted GitLab AI Gateway environment. The reported exploit path requires an authenticated account with Duo Agent Platform access. The listed fixes are AI Gateway versions 19.2.4, 19.3.2 and 19.4.1, but administrators should confirm the applicable fixed version in GitLab’s current security advisory before treating a deployment as patched.
What is CVE-2026-90970?
Cyber Threat Observatory describes CVE-2026-90970 as improper neutralization of special elements in a template engine (CWE-1336). In its account, a crafted flow configuration can escape the template sandbox and execute arbitrary commands in the AI Gateway environment. The report assigns the issue a CVSS 3.1 score of 9.9 (Critical); that is the report’s assessment, not a rating independently confirmed in a GitLab advisory.
The reported impact is command execution in the gateway environment, not an established compromise of every GitLab installation. The report concerns self-hosted AI Gateway deployments and says exploitation requires an authenticated user with Duo Agent Platform access. These vulnerability-specific details, including the reported status of GitLab-managed gateways, come from the secondary report because the linked GitLab work item does not display the advisory details. See the Cyber Threat Observatory report and check GitLab’s current advisory for confirmation.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Am I affected?
Start by establishing whether your organization runs a self-hosted AI Gateway, then check its deployed image or Helm release. GitLab documents AI Gateway installation methods, including Docker and Helm, and version-tagged images.
#1 Best Overall
The report lists these affected ranges and fixed versions. Because the precise boundaries are secondary-source claims, use the table as an inventory prompt, not as a substitute for GitLab’s current advisory.
| Reported deployment range | Reported fixed version |
|---|---|
| 18.1.6 through 19.2.4, according to the report | 19.2.4 |
| 19.3 before 19.3.2, according to the report | 19.3.2 |
| 19.4 before 19.4.1, according to the report | 19.4.1 |
The range wording above follows the technical report and should not be read as an independently validated version boundary. Compare the exact image tag or Helm release with the GitLab security advisory and GitLab’s security guidance. The report says GitLab-managed gateways were already patched; confirm the current status with GitLab rather than assuming that statement applies to every hosted configuration.
How should administrators respond?
- Inventory the gateway. Determine whether the deployment is self-hosted and record the deployed container image tag or Helm release.
- Verify the advisory. Check GitLab’s current security advisory and supported-release guidance for the exact affected and fixed versions relevant to your installation. GitLab’s security FAQ recommends upgrading to at least the latest security release for the supported version and notes that ad-hoc security releases may be issued for critical vulnerabilities.
- Upgrade if required. Use the supported upgrade path for your Docker or Helm deployment. Confirm the running gateway reports the intended image or release after rollout; do not infer that a deployment is fixed solely because its version appears in a secondary report.
- Review access. Identify accounts with Duo Agent Platform access and follow your organization’s access-control policy while remediation is underway. The report describes this access as a prerequisite for the reported exploit path.
- Restrict outbound traffic carefully. GitLab recommends limiting traffic from the AI Gateway container to necessary destinations, including the GitLab instance, configured model-provider endpoints and, where applicable, GitLab license validation. Test firewall rules before production rollout: rules that are too restrictive can interrupt gateway functionality.
What is confirmed—and what needs checking?
GitLab’s public installation guidance supports the operational steps of identifying the gateway deployment and controlling its outbound connections. GitLab’s security FAQ provides general upgrade advice. However, the affected-version boundaries, CVSS score, exploit prerequisites and GitLab-managed gateway patch status cited here are reported by Cyber Threat Observatory, not verified from a rendered GitLab advisory. Treat those details as provisional until checked against GitLab’s current notice.
Quick Recap
Best Value
- Used Book in Good Condition
Rank #4
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
