GitLab 19.4 Adds Goal-Driven Agents and Tighter Cost Controls

GitLab has released version 19.4 of its DevSecOps platform, adding a goal-driven mode to GitLab Duo CLI, three hosted open-weight models, a larger set of Model Context Protocol (MCP) tools, and more detailed controls for AI spending. The release moves GitLab's agents beyond individual prompts and predefined flows, while keeping their actions within the permissions and approval rules already used by a project.

The new /goal command in GitLab Duo CLI, now in public beta, accepts an open-ended objective and continues working until it completes the task or reaches an iteration limit. One model implements the changes while a separate model checks progress against the stated goal. A developer can stop the run, change the goal, or redirect the agent. The work runs locally and retains the project's existing context and controls.

"This release takes agentic automation from something individual developers use to something an organization can scale at speed and under the controls already in place."
Manav Khurana, GitLab

GitLab's use of the word "verified" has a specific meaning here. The second model judges whether the result meets the natural-language goal; this is not formal verification of the code. The command can report work that it could not resolve, and the iteration limit prevents it from continuing indefinitely. Clear acceptance criteria therefore remain important, even though the developer no longer has to supervise every intermediate step.

GitLab Duo Agent Platform also gains support for the hosted open-weight models Kimi K3, MiniMax M3 and GLM 5.3. GitLab says these provide up to four times as many calls per GitLab Credit as many comparable frontier models. Group owners can set a default model for a feature and restrict the models available to teams. This lets an organisation use a lower-cost model for routine work and reserve a more expensive model for tasks where it performs better. The announcement does not include model-by-model benchmark results or absolute per-call prices, so customers will still need to test the options against their own workloads.

"The platform running the automation is what governs which tools an agent can touch and attributes what it consumes."
Manav Khurana, GitLab

The expanded GitLab MCP server tools let agents in third-party clients trigger and inspect pipelines, open and review merge requests, update work items, browse repositories, and manage vulnerabilities. Read-only tools default to "Always Allow", while write and delete operations default to "Always Ask". Administrators can change those rules at group or project level. Most of the MCP tools are available on the Free, Premium and Ultimate tiers, although the vulnerability management tools require Ultimate.

This is part of a wider move to let coding agents act through MCP without granting unrestricted access. InfoQ reported in 2025 that Google's Agent Mode for Android Studio could use MCP to reach external tools. GitHub Copilot also supports MCP across its main interfaces, with enterprise policies for controlling access to servers. GitLab's approach applies a decision to each tool and places a human checkpoint in front of changes by default.

Cost controls matter more when agents run automatically rather than in response to an individual prompt. DevOps consultancy and GitLab partner ReleaseTEAM cautions that "the practical risk is automation", because a flow attached to every pipeline failure or new merge request could run hundreds of times a day. It recommends setting usage caps before connecting a flow to a high-volume trigger, then measuring consumption in a limited pilot.

The fourth main change concerns cost attribution. Credit usage exports now include a row for each billable event, with the user, project, session, flow type, credits and token counts. A new settings page lets owners set a default per-user cap and add individual overrides without using the GraphQL API. Developers can see their own consumption, while billing managers receive warnings when on-demand use reaches 50% or 80% of a monthly cap.

The release follows a series of changes to GitLab Duo during 2026. GitLab 19.2 put agents to work on dependency remediation and security reviews, with a person still responsible for approving changes. Earlier, versions 18.10 and 18.11 introduced flat-rate automated code reviews, free-tier access to the Duo Agent Platform, and monthly credit limits. Version 19.4 extends that approach to longer-running development tasks and gives administrators a closer view of the resulting costs.

GitLab 19.4 was released on 17 September 2026. The /goal command is available to Premium and Ultimate customers on GitLab.com, GitLab Self-Managed and GitLab Dedicated, and requires GitLab 19.3 or later with Duo CLI 9.17.0 or later. The command and expanded MCP tools remain in public beta, while the new credit usage visibility is generally available.