GitLab has released version 19.3 of its platform, introducing updates designed to enhance control and security for enterprises adopting agentic software development. Key features include the integration of the GitLab Duo Agent Platform into GitLab Dedicated environments, a new Secrets Manager, and tools for bulk SAST vulnerability remediation.
For GitLab Dedicated customers, the AI Gateway for GitLab Duo Agent Platform now operates within their existing single-tenant SaaS infrastructure. This allows agentic workloads to adhere to the same data residency and isolation models as other software development lifecycle components, enabling organizations to connect their own models for inference and maintain AI-processed data within their established security boundaries.
The new GitLab Secrets Manager is now available in limited release as a paid add-on for GitLab.com customers, billed via GitLab Credits. This feature extends secret management beyond pipelines, ensuring that every CI secret is scoped to the specific environment, branch, and protection status of the job requiring it. The manager also supports Kubernetes, Terraform, OpenTofu, and custom tools, aiming to consolidate credential management within the same platform that runs code and pipelines.
GitLab 19.3 also introduces capabilities for bulk SAST false positive detection and agentic SAST vulnerability resolution. These tools allow security teams to address multiple findings in the Vulnerability Report simultaneously. The system provides a confidence score for each finding, and confirmed risks can generate ready-to-merge fixes, streamlining the remediation process for entire vulnerability backlogs. GitLab will continue to automatically triage and remediate new critical and high-severity findings as they emerge.
The Flow Creator Agent, accessible through Agentic Chat within the GitLab Duo Agent Platform, aims to simplify custom automation. Users can describe desired automations in plain language, and the agent generates a runnable flow ready for registration from the AI Catalog. These flows operate under a scoped service account with composite identity, requiring a Maintainer role or higher to enable.
Additional features in GitLab 19.3 include the general availability of GitLab Credits usage caps, allowing administrators to set monthly spending limits for agentic AI. These caps can be configured at the subscription level in the Customers Portal, with options for default per-user caps or per-user overrides via the GraphQL API. Restricted visibility for custom agents and flows per GitLab group is also now generally available, making them accessible to members across multiple projects within a group, in addition to existing per-project and public visibility options.
