# GitLab Unveils Agent Orchestration to Automate End-to-End DevOps Delivery New goal-driven flows and Slack integration eliminate downstream manual bottlenecks across software verification. - **Source:** [GitLab](https://brifing.si/source/76101668-80db-4240-8676-e968d2b72546) - **Published:** 2026-10-13 - **Topic:** [Software & Engineering](https://brifing.si/c/engineering) - **Original article:** - **On Brifing:** https://brifing.si/s/eed15d6f-77f5-431d-a817-e6cd2ecf3df9 ## Resolving Downstream DevOps Bottlenecks AI code generators accelerate initial code writing, but software delivery remains stalled during downstream phases. Stages like code review, pipeline verification, scan triage, and manual approvals still run on human time. GitLab announced agent orchestration capabilities in Duo Agent Platform to bridge these fragmented, manual handoffs across the lifecycle. ## Goal-Driven Autonomous Sequences Goal-driven flows allow engineers to define high-level outcomes rather than step-by-step prompts. Triggered via the slash goal command, flows autonomously progress through tests, reviews, and security scans until requirements are met. An independent model evaluates progress at each step, allowing developers to pause, redirect, or inspect execution at any stage. ## Execution Across Interfaces Developers can invoke goal-driven flows across multiple operational environments, including Agentic Chat, Duo CLI, and headless execution via CLI commands. This flexibility enables engineers to integrate autonomous remediation tasks directly into terminal-based developer environments, local continuous integration loops, or scriptable automated systems without relying solely on traditional web graphical interfaces. ## Direct Slack Integration Workflows GitLab for Slack enables teams to initiate engineering workflows directly from communication threads without context switching. Mentioning the bot triggers pipeline fixes, creates trackable issues, or answers codebase questions. Progress streams in real time back into the Slack channel while maintaining the developer identity, enterprise policy, and audit trail. ## Custom Multi-Step Event Flows Custom Flows chain specialized AI agents together to mirror existing internal processes. Triggered by platform events like user mentions, issue assignments, or pipeline failures, they automate complex operations. For instance, self-healing pipelines analyze test failures, generate targeted code patches, commit changes, and request human review at sensitive checkpoints. ## Extensibility via MCP Architecture The Duo Agent Platform supports external extensibility by connecting external agents and internal tools through Model Context Protocol client and server implementations. Engineering teams can leverage pre-built foundational agents from the AI Catalog, orchestrate proprietary tooling, and bind external services into unified, auditable delivery pipelines across teams. ## Governance and Evidence Chains All automated agent actions execute strictly within GitLab identity boundaries, established role permissions, and formal compliance evidence chains. Because agents run under verified enterprise governance rather than unmonitored external contexts, security policies and compliance oversight automatically accompany every automated code commit, security scan, and pipeline execution without compromise. ## Availability and Version Requirements Goal-driven flows require GitLab version 19.3 or later alongside Duo CLI 9.17.0 on GitLab.com, Self-Managed, or Dedicated instances. Access is provided for Premium and Ultimate tiers utilizing subscription credits. Custom Flows are generally available, while MCP server integration and Slack tools launch generally later this month. ## Sources covering this story - https://about.gitlab.com/blog/gitlab-agent-orchestration/ — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._