
GitLab 19.6 expands AI agents across security work and Slack
The AMW Read
The release extends an established developer platform's agents into security and operations workflows, with implications for competition across developer tools.
GitLab 19.6 expands AI agents across security work and Slack
GitLab 19.6 adds tools for security teams to run GitLab Duo agent flows across multiple projects and non-default branches. A beta flow adds exploitability context to static application security testing findings. On GitLab.com, a separate beta can revoke an exposed GitLab personal access token, create a replacement with the same permissions, store it in GitLab Secrets Manager and open a merge request to remove the hardcoded token.
The release expands GitLab's AI coding tools beyond code generation into security remediation, recurring operations and team conversations. Scheduled agent flows are in beta and can run from every 15 minutes to every three months using service accounts with scoped permissions. GitLab for Slack is generally available on GitLab.com for Premium and Ultimate customers. It lets users request issues, merge requests or agent flows from a Slack thread or direct message, subject to their GitLab permissions.
For engineering teams, the practical test is whether these flows reduce backlog and triage time while keeping changes reviewable. GitLab says Slack-initiated merge requests follow existing reviews, approvals and audit trails. It also says most customers will see their credits go up to twice as far because tasks require fewer model requests. The new credit terms took effect November 19, 2026, for new customers; existing customers move on December 18 or can opt in earlier.