Two releases shipped back to back this week and together they push Kubernetes Sentinel into genuinely new territory.

v1.5.0: Three Features Nobody Has Built Into a Lightweight Kubernetes Tool

The first is natural language kubectl. You type what you want to do in plain English directly in the dashboard. The active LLM translates it into the exact kubectl command, scores it as low, medium, or high risk with a plain English reason, and gives you a Run button. No more memorizing flags. No more checking documentation at 2am during an incident.

The second is cluster conversation. There is now a full persistent chat interface in the dashboard where you talk to your cluster like a colleague. Ask it anything. Which pod has the most restarts right now? What happened in the last hour? Why is my ml-pipeline failing? The LLM has complete live context of every pod, every event, every node, and every deployment across every namespace and answers from real data, not guesses. Four starter questions guide first-time users and the conversation persists until you clear it.

The third is the incident report generator. One click produces a full professional postmortem document. It pulls the health timeline from the SQLite database, the AI diagnosis history, and the current cluster state, and generates a structured report with a summary, a timeline table, an impact assessment, the root cause, remediation steps, and three specific prevention recommendations based on the actual failure pattern. Download it as a markdown file or copy it straight to Confluence or Notion. Engineers spend one to two hours writing these after every incident. Now it takes ten seconds.

The dashboard is now at nine tabs. kubectl, Chat, and Report sit alongside the existing Pods, Events, Resources, Timeline, History, and Diagnosis tabs.

v1.6.0: Automated CI/CD and One-Command Deployment

Version 1.6.0 is the infrastructure release that makes the project genuinely usable by anyone without any setup friction.

GitHub Actions now automatically builds and publishes the Docker image to GitHub Container Registry on every push to main. The image is tagged with the version number, the major.minor version, and latest. Multi-architecture builds support both amd64 and arm64 so the sentinel runs on any cluster regardless of node architecture.

A second CI workflow runs on every push and pull request, checking Python syntax across all backend files and validating that all imports resolve correctly. Pull request builds run the full build without pushing so contributors can verify their changes are safe before merging.

The practical outcome is that deploying Kubernetes Sentinel to any cluster now takes five kubectl commands and thirty seconds. No cloning the repo. No installing Python. No building a Docker image locally. Just apply the manifests directly from the raw GitHub URLs and the published image pulls automatically.

The project is at github.com/OsamaOracle/k8s-sentinel.