A Deep Dive into the Agent, Cluster Access and Security Features
summary
Explains how GitLab Agent installs into a Kubernetes cluster as a minimal client while moving control logic to the GitLab server, communicating over a bidirectional gRPC stream so the cluster does not need to expose the kube-apiserver. Shows a working demo on a Kind cluster with instant token-based registration via a three-step Helm install, traces deprecations (integrated cluster UI deprecated in 14.5; move to GitLab Agent + Flux from 16.2), and covers access controls including Developer+ gating, user impersonation, Trivy-based monitoring, and restricting the Agent with a serviceAccount.
AI-written summary · about
chapters (11)
- 0:00GitLab and Kubernetes
- 7:24GitLab Agent
- 10:35Agent design benefits
- 12:44CI/CD with Agent
- 13:50GitOps and Flux
- 17:02Agent for deployment
- 19:59Kubernetes access via GitLab
- 21:25Agent access demo
- 28:40Impersonation and permissions
- 29:43Monitoring and Agent future
- 30:27Q&A
related talks
22mKubernetes Environments Visualisation and How We Do It
23mAgents in Kubernetes 101: A Crash Course on Cloud Native AI Agents
27mBeyond Chat: Bringing Agents To Kubernetes
22mGitLab + Flux!
15mPlatform Engineering Done Right: Safe, Secure, & Scalable Multi-Tenant GitOps
30mSo You Want To Run AI Agents on Kubernetes: A 101 Guide
33mKubernetes As A Service with GitOps
14mFrom the Ranch(er) to the Dev: A Case Study About Simplifying K8s Cluster Setup Wit...