GitLab 19.5, released on October 15, makes the GitLab MCP server generally available and introduces governance capabilities for external AI agents.
The release brings agentic development to the forefront of the platform, allowing MCP-compatible tools such as Claude Code and Cursor to access GitLab's broader tool catalogue and invoke GitLab Duo Agent Platform agents and flows. The GitLab MCP server, now generally available, provides agents with access to capabilities through the Model Context Protocol. New tools let agents discover groups, inspect individual CI/CD artifact files, and start or monitor GitLab Duo agents and flows. Administrators can select which MCP toolsets are exposed, reducing unnecessary context consumption and providing a relatively simple form of least-privilege agent design.
A significant addition is the ability to bring Externally Connected Agents into GitLab's AI Governance capabilities. Previously focused on GitLab Duo Agent Platform agents, the governance dashboard can now support organisations in recording qualifying sessions from tools like Claude Code, Cursor, and Codex when connected through the GitLab CLI. Activity appears alongside GitLab's own agents, and audit events can be streamed to a SIEM, giving security teams visibility into which agents are being used, by whom, and against which projects.
GitLab Duo CLI gains an auto mode, allowing agents to execute tools without requesting approval for every shell command, Git operation, or MCP call. The feature is disabled by default and can be controlled at instance, group, and project levels, while tools marked Always Deny remain blocked. The /goal command is also now generally available, letting developers describe an open-ended objective, with GitLab Duo delegating its implementation and verification to a governed flow.
GitLab Secrets Manager is now generally available. Built on OpenBao, it integrates with existing GitLab group and project permissions, scopes secrets to the jobs that require them, and generates audit events for secret operations. Security configuration profiles have also been expanded, allowing organisations to centrally customise settings for Secret Detection and SAST rather than repeating configuration across every project's CI/CD file.
The release follows a similar trajectory to GitHub's strategy of integrating coding agents into the software lifecycle, but GitLab emphasises connecting external agents to its existing DevSecOps governance model. The competitive question is moving beyond which coding assistant generates better code to which platform provides the safest environment for autonomous agents to perform engineering work.
Analysis
Why This Matters
- Development teams gain a governed path to adopt AI agents without sacrificing security or compliance oversight.
- Organisations can centralise control over external coding agents, reducing the risk of unauthorised or unmonitored agent activity in production pipelines.
- The release signals a shift in the DevOps platform war: from code generation quality to safe, auditable agent orchestration.
Background
GitLab 19.5 is the latest version of the company's DevOps platform, which provides source code management, CI/CD, security scanning, and monitoring in a single application. Agentic development refers to AI agents that can autonomously perform multi-step engineering tasks, such as writing code, running tests, and managing deployments. Prior to this release, GitLab's AI governance focused solely on its own GitLab Duo agents.
What to Watch
- Adoption rate of the MCP server and external agent governance features among enterprise customers.
- How competitors, particularly GitHub, respond with similar governance capabilities for third-party agents.
- Security teams' reception of the new audit and SIEM streaming capabilities, and whether any gaps emerge in agent activity tracking.
Checked
12 claims in this story checked against 1 source: 12 supported.
Show each claim
- SupportedGitLab 19.5 puts agentic development at the centre of its DevOps platform.InfoQ: “puts agentic software development and governance at the centre of the platform”
- SupportedGitLab 19.5 makes the GitLab MCP server generally available.InfoQ: “making its GitLab MCP server generally available”
- SupportedGitLab 19.5 adds governance capabilities for external AI agents.InfoQ: “GitLab is adding controls for governing those external agents”
- SupportedGitLab 19.5 was released on October 15.InfoQ: “GitLab 19.5 , released on October 15, puts agentic software development and governance at the centre of the platform”
- SupportedMCP-compatible tools such as Claude Code and Cursor can access GitLab's broader tool catalogue and invoke GitLab Duo Agent Platform agents and flows.InfoQ: “The release allows MCP-compatible tools such as Claude Code and Cursor to access GitLab's broader tool catalogue and invoke GitLab Duo Agent Platform agents and flows.”
- SupportedThe MCP server adds tools to discover groups, inspect individual CI/CD artifact files, and start or monitor GitLab Duo agents and flows; administrators can choose which toolsets are exposed.InfoQ: “New tools allow agents to discover groups, inspect individual CI/CD artifact files, and start or monitor GitLab Duo agents and flows. Administrators can also select which MCP toolsets are exposed”
- SupportedClaude Code, Cursor, and Codex can connect through the GitLab CLI so organisations can record qualifying sessions, with audit events streamable to a SIEM.InfoQ: “GitLab 19.5 allows tools such as Claude Code, Cursor, and Codex to be connected through the GitLab CLI. Once configured, organisations can record qualifying agent sessions”
- SupportedGitLab Duo CLI auto mode lets agents execute tools without approval for every shell command, Git operation, or MCP call; it is disabled by default, controllable at instance, group, and project levels, and does not allow tools marked Always Deny.InfoQ: “The feature is disabled by default and can be controlled at instance, group, and project levels, while tools explicitly marked Always Deny remain blocked.”
- SupportedThe /goal command is generally available and lets developers set an open-ended objective for GitLab Duo to delegate implementation and verification to a governed flow.InfoQ: “The /goal command is also now generally available, allowing developers to describe an open-ended objective and have GitLab Duo delegate implementation and verification to a governed flow.”
- SupportedGitLab Secrets Manager is generally available, is built on OpenBao, uses existing group and project permissions, scopes secrets to required jobs, and generates audit events for secret operations.InfoQ: “Built on OpenBao , it integrates with existing GitLab group and project permissions, scopes secrets to the jobs that require them, and generates audit events for secret operations.”
- SupportedSecurity configuration profiles let organisations centrally customise Secret Detection and SAST settings rather than repeating configuration in every project's CI/CD file.InfoQ: “allowing organisations to centrally customise settings such as excluded paths, analyzer images and full-history scans for Secret Detection and SAST rather than repeating configuration in every project's CI/CD file.”
- SupportedGitLab's approach resembles GitHub's strategy of integrating coding agents into the software lifecycle, while emphasising governance for external agents; the competitive question is which platform offers a safe environment for autonomous engineering work.InfoQ: “GitLab's direction increasingly resembles GitHub's strategy of turning coding agents into participants in the software lifecycle , but GitLab is placing particular emphasis on connecting external agents to its existing DevSecOps governance model.”
