GitLab 19.6 adds agentic remediation and Slack workflows
GitLab 19.6 adds cross-project agent remediation, token rotation, scheduled agent flows and Slack-triggered work. GitLab says revised credits stretch up to twice as far for most customers.

Why it matters
- GitLab released version 19.6 on November 20, 2026.
- Scheduled automations can run every 15 minutes to every three months.
- Rotate-and-fix revokes a leaked token, creates a replacement and opens a merge request.
- Most customers can stretch GitLab Credits up to twice as far.
- Rapid Diffs loads the first file about three times faster and is generally available.
By Rebecca Stone
GitLab 19.6, released November 20, 2026, expands agentic security remediation, recurring workflows and Slack-triggered development across the company's DevSecOps platform.
GitLab said the release targets software, security and product teams that want software agents to handle routine and remedial work while people retain oversight of what those agents do and what they cost.
GitLab is placing agents at three operational points: vulnerability remediation, recurring background jobs and requests made inside workplace messaging. The controls around those actions come from the same release, including scoped service accounts, designated approvers, merge-request review and audit events.
What does agentic remediation change?
Security managers can now enable agent flows across multiple projects at once, according to GitLab. Development teams can run those flows on branches other than the default branch, allowing agents to address findings before code merges.
A vulnerability context flow, released in beta for SAST findings, is designed to show whether a flaw is exploitable in practice. GitLab said teams can rank issues by exploitability as well as severity, then decide which findings to fix first.
GitLab's intended path starts with a finding, adds exploitability context, routes an agent to a branch and applies an approval policy where a change contains a critical or known-exploited vulnerability. The company said the point is to remediate before merge rather than leave every fix to a developer.
The release also adds a mean time to remediation chart. Group-level policies can hold changes containing critical or known-exploited vulnerabilities until a designated approver signs off. GitLab said those controls are meant to preserve human oversight as remediation expands across projects.
How does token rotation work?
The credential response combines secret detection with code repair. When secret detection finds an active GitLab personal access token, teams can use a beta rotate-and-fix function on GitLab.com to revoke it and remediate the code in one step.
GitLab says the function revokes the leaked token, creates a replacement with the same permissions, stores the replacement in GitLab Secrets Manager and opens a merge request that removes the hardcoded token from source code.
The sequence covers both the live credential and its copy in source code. GitLab presents the combination as a one-step path between a secret detection and a merge request that removes the exposed value.
Manav Khurana, Chief Product and Marketing Officer at GitLab, said: "Every organisation is racing to keep up with the security challenges that have become the new norm. GitLab 19.6 helps security teams clear vulnerability backlogs faster, and provides new automations. Both of which are easier to afford now that each GitLab Credit can be used for more agentic work."
Can agents run on a schedule?
GitLab is also introducing scheduled automations in beta for GitLab Duo Agent Platform flows. The feature runs those flows in the background on a recurring schedule, with intervals ranging from every 15 minutes to every three months.
GitLab listed daily issue triage and hourly checks on CI health as example jobs. Each automation runs through a dedicated service account with scoped permissions, while audit events record the automation separately from human actions.
GitLab designed the schedule and permission layers to work together. The interval determines when a flow runs; the service account determines what it can access; the audit record identifies the run as machine activity rather than a human action.
The redesigned Automations page is intended to answer three operational questions:
- What ran.
- How often a run succeeded.
- Why a run failed.
GitLab said the view is intended to make recurring agent-driven work easier to monitor and govern.
How does Slack fit into GitLab's review process?
GitLab for Slack is now generally available on GitLab.com for Premium and Ultimate customers. Users can mention @GitLab in a Slack thread or direct message to create issues, merge requests and agent flows, with results posted back to the thread.
According to GitLab, each request runs with the user's own GitLab permissions and creates an agent session inside GitLab. Merge requests created through Slack still go through the usual review, approval and audit processes.
The request therefore stays connected to GitLab's identity and workflow controls. The Slack thread receives the outcome, but the underlying issue, agent session or merge request remains visible inside GitLab and follows its established process.
Slack said the integration reflects a software workflow in which many decisions begin in workplace messaging before engineers enter them into engineering systems.
Katie Steigman, Vice President of Product Management at Slack, said: "Teams make decisions in Slack every day, and the best agents are the ones that can act on that context. GitLab for Slack lets agents read the conversation, take action across the software lifecycle in GitLab, and bring the results back to the thread. Teams can go from a quick discussion to planned, reviewed, and shipped work without switching tools."
What changes about GitLab Credits?
GitLab has also changed the cost equation for Duo Agent Platform usage. The company said it made that usage more efficient, so each task now requires fewer model requests and most customers will see their GitLab Credits stretch up to twice as far.
GitLab tied the revised economics to heavier use of agentic features, including remediation flows, scheduled automations and Slack-based actions. Customers on Flex contracts can direct credits toward other products, including GitLab Secrets Manager and Dedicated Hosted Runners.
Scheduled automations already allow agent flows to run in the background, and the other new features add more places for agents to act. GitLab said the new credit efficiency is intended to support heavier use of those features, including remediation flows, scheduled automations and Slack-based actions.
Flex contract customers can direct credits toward GitLab Secrets Manager and Dedicated Hosted Runners in addition to Duo Agent Platform usage. That gives buyers more flexibility in allocating the balance they use for agentic work.
Which governance and performance controls arrive?
Beyond the headline workflows, GitLab 19.6 adds MCP governance controls in beta. Settings can mark specific tools to always allow, always ask or always deny, and GitLab creates an audit event for every tool call.
The release pairs those controls with flow suggestions when external MCP clients connect. It also makes a Rapid Diffs view generally available, which GitLab said loads the first file about three times faster.
GitLab also supports resumable jobs and auto-merge time windows. The company said the broader release extends its push to make agent-based software work routine inside development and security workflows while preserving approval checks and activity records.
What should customers watch next?
GitLab 19.6 marks vulnerability context, rotate-and-fix, scheduled automations and MCP governance as beta. The next test for GitLab is whether teams adopt those workflows in daily security and development work while keeping the approval gates and audit trails intact.
Source: itbrief.com.au
More from Rebecca Stone
Show full bio
Correspondent covering companies, deals and strategy in artificial intelligence and machine learning at AI In Context.
802 articles