GitLab 19.3: Securely Scale Agentic Development for Enterprises

Post Views: 12

GitLab 19.3 introduces enhancements to support secure scaling of agentic software development for enterprises.

Key Features

The update includes several key features: Secrets Manager, Flow Creator Agent, Bulk SAST False Positive Detection, and Agentic SAST Vulnerability Resolution.

Secrets Manager
Flow Creator Agent
Bulk SAST False Positive Detection
Agentic SAST Vulnerability Resolution

AI Gateway for GitLab Duo Agent Platform

The AI Gateway for GitLab Duo Agent Platform now operates within GitLab Dedicated’s single-tenant SaaS infrastructure, aligning agentic workloads with the same residency and isolation standards as other software delivery processes. This enables regulated teams to leverage agentic AI for development tasks under the same deployment framework auditors recognize.

GitLab Secrets Manager

GitLab Secrets Manager, now in limited availability as a paid add-on for GitLab.com users, scopes CI secrets to specific environments, branches, and job protection levels. It also supports Kubernetes, Terraform, OpenTofu, and custom tools, consolidating credential management within the platform that hosts code and pipelines.

Bulk SAST Remediation

Bulk SAST remediation capabilities allow teams to address years of accumulated risk through automated processes. SAST False Positive Detection and Agentic SAST Vulnerability Resolution enable simultaneous resolution of multiple vulnerabilities in the Vulnerability Report. Each finding receives a confidence score, with confirmed risks generating ready-to-merge fixes for developer review.

Flow Creator Agent

The Flow Creator Agent eliminates barriers to custom automation by removing the need for manual schema mapping. Available via Agentic Chat in the GitLab Duo Agent Platform, it converts plain-language automation requests into executable workflows.

Additional Features

Additional features in GitLab 19.3 include general availability of GitLab Credits usage caps, allowing organizations to set monthly limits on agentic AI expenditures. Administrators can define subscription-level caps through the Customers Portal, with options for per-user limits or overrides via the GraphQL API. Restricted visibility settings for custom agents and flows within GitLab groups are now generally available, limiting access to specific projects within a group. This complements existing per-project and public visibility options.