GitLab Duo Agent Platform

Higher agent accuracy, lower token cost
Learn how to harden your software factory and let AI agents earn autonomy, without trading away security.
Stay updated with our latest features and improvements.
Specialized agents, customizable for your work
Move faster with expert-built agents your team can use out of the box and extend with custom versions to reflect your organization's standards and workflows.
Invoke agents designed for planning, coding, security analysis, analytics, and more. Each agent uses GitLab context to deliver precise, relevant results that support real work.
Give teams a centralized place to explore, activate, and manage agents and flows. With the AI Catalog, new automation becomes simple to adopt and consistent to operationalize across projects and groups.
Create or refine agents so they follow your conventions, mirror your quality expectations, and support the workflows that matter to your teams.
Integrate external agents such as Claude Code and Codex into your GitLab workflow.
Automate complex tasks with agentic workflows
Coding agents get you to the MR. GitLab Duo Agent Platform is the repo-side AI that carries the change through review, tests, security, and pipelines, to a change that actually ships.
Duo Agent Platform picks up where coding agents stop, handling the review, security, and pipeline steps.
Kick off automation from a GitLab event, a recurring schedule, or the tools your team already works in, so it runs without anyone starting it by hand.
Run automation under your own rules and identity, the same way on every project, and put a human checkpoint where your policy calls for one.
See how GitLab Duo Agent Platform automates software delivery with AI agents
Enterprise-grade governance and control
Get what your organization needs to use AI responsibly and at scale, with clear permissions, boundaries and transparency built in.
Decide which agents and flows are allowed, where they can operate, and which models they can use. Everything aligns to your existing roles and group structure.
Trace every agent action back to a real user's identity and permissions, with sessions, tools, models, and approvals all in one trail.
Run self-hosted models on GitLab Self-Managed, or bring your own model into a single-tenant AI Gateway on GitLab Dedicated, under one consistent policy.
Extensible and flexible
Duo Agent Platform is designed to plug into the way your organization already operates so the platform adapts as your environment evolves, not the other way around.
Connect your tools to GitLab over MCP so agents pull real context. Or call into GitLab from the coding agents you already use.
Agents and flows can be updated, extended, or replaced over time, helping teams evolve their AI strategy while preserving consistency across the platform.
Run your agents, flows, and agentic chat on the right LLM for the job, from a roster you control.
GitLab Orbit is the lifecycle context graph that natively supports GitLab Duo Agent Platform and your software development agents. It connects your code, pipelines, and deployments as grounded context for every agent.
From specialized agents that understand your security posture to automated flows that eliminate manual handoffs between software lifecycle stages, here's a look at what's possible with GitLab Duo Agent Platform.
Transform ideas into structured plans, identify stale backlog items, and more.
Analyze vulnerabilities, dismiss false positives, and get security guidance.
Gather feedback, iterate on merge requests, and more.
Analyze the root cause and generate a fix in a new merge request.
Automatically migrate CI/CD configurations and reduce manual effort.
Automatically transform issues into production-ready code.
Create, test, and share custom agents with specific skills.
Chain agents together to automate complex tasks.
Scale use of agentic AI across your projects and team members.
Accelerate development, improve code quality, and automate documentation.
Add new features, fix existing ones, and modernize code.
Get actionable guidance across issues, comments, and more.
We've created a maturity framework based on our experience working with industry-leading customers. It offers prescriptive guidance on how to rapidly progress through three maturity stages.