GitLab

GPT-5.6 (Sol, Terra, & Luna) is now live in GitLab Duo Agent Platform, giving you another option for your agentic workflows. Try it today.

Your agents are working. But are they working for you? Take back control with GitLab. https://lnkd.in/e6rCQWN6

How does GitLab make sure its own security detections work? Meet WATCH, formally known as Oops, a project built by our Signals Engineering team using GitLab Duo Agent Platform. The team is responsible for catching malicious behavior across GitLab's infrastructure, while making sure legitimate, everyday Git work never trips the alarm. Security engineer Evan Baltman built a custom Duo agent skill that writes Python unit tests simulating malicious behavior, like a webhook being added to exfiltrate project data. If the detection fires, it's working. If it doesn't, the team gets a Slack alert before a real gap becomes a real problem. It's a great look at Duo Agent Platform doing real security work: writing tests, using context from past examples, and helping a small team keep hundreds of custom detections reliable.

Your CI/CD variables were never meant to hold your most sensitive credentials 👀 In this episode of Dev Snacks, Staff Developer Advocate Itzik Gan-Baruch walks through the new GitLab Secrets Manager and why storing secrets in standard variables introduces risk your team doesn't need to carry. He shows how a built-in secrets manager keeps workflows moving while scoping access strictly to the execution runtime.

Looking for the one? We’re talking about your workflow, of course 👀 Swipe through to meet the GitLab AI agents helping teams plan, code, and secure with a whole lot more flow. Get to know them better in the GitLab Duo platform 👉 https://lnkd.in/ejU-Dz38

AI is about to change the vulnerability landscape faster than most security teams are ready for. GitLab's Jamie Dicken talks to SC Media about the coming surge in AI-discovered vulnerabilities, why context is the missing piece for AI agents in the SDLC, and how knowledge graphs like GitLab Orbit give agents an authoritative map instead of a guess. She also breaks down what real governance looks like, pairing human oversight with authorization aware access and clear guardrails so speed doesn't turn into chaos. Read the full conversation to see how GitLab is preparing teams for the scale of AI-accelerated offense. https://lnkd.in/eZbQX-yG

AI made coding faster. Now source code management needs to keep up. At GitLab Transcend in London, we introduced our next-generation Source Code Management architecture: ⚡ Up to 42x faster ⚡ Less waiting ⚡ More shipping Watch the demo to see how we're building for the agentic era. https://lnkd.in/gUHtshkx

Learn how GitLab's restricted access feature prevents seat overages, integrates with your identity provider, and keeps license costs predictable. https://lnkd.in/eNhA48KC

What happens when AI generates more code than your organization is prepared to govern? That's the question at the center of this month's Monday Merge. https://lnkd.in/ePnCczC3 Our new AI Accountability Report surveyed more than 1,500 DevSecOps professionals and found a gap that should worry every engineering leader: 78% say developers are shipping code faster with AI, but 82% believe it's creating a new form of technical debt they're not ready to manage. And 80% admit they adopted AI tools faster than they built the policies to govern them. This is exactly why GitLab 19.1 matters. The new One Vulnerability View brings scanner coverage, vulnerability management, and AI governance into a single place, so teams can see across third-party scanners, stream AI audit events, and apply approval guardrails to agent actions. Catch up on the newsletter, grab the report, and see us at our upcoming events!

AI coding has moved past experimentation. Now the real question is whether software delivery can keep up. Our new AI Accountability Report surveyed 1,528 DevSecOps professionals and found 79% say developer productivity has improved, but the broader delivery lifecycle hasn't kept pace. The bottleneck hasn't disappeared, it's just moved from writing code to reviewing, validating, and securing it. This month's Monday Merge digs into what that means and what we're building in response: GitLab 19.1's new governance guardrails for AI agents, Next-Gen Source Code Management (up to 50x faster task execution for agents), and GitLab Orbit, our new context graph now in public beta. Read the full edition here: