Private beta, by invitation

Krufter — the AI-assisted Kubernetes GUI

Krufter is a desktop application to explore, diagnose and operate Kubernetes clusters, with a built-in AI assistant that explains errors and proposes fixes without exposing your secrets.

What is Krufter?

Krufter is a desktop Kubernetes client for Linux, macOS and Windows, in the spirit of Lens: it connects straight to your clusters from your kubeconfig, locally, and presents your resources (pods, deployments, namespaces, nodes, events) in a fast, keyboard-first interface.

It adds an AI assistant (Google Gemini) that understands the context of a failing resource, explains the probable cause and proposes a fix.

Who is it for?

Developers, DevOps / SRE engineers and platform teams who run Kubernetes clusters every day and lose time chaining kubectl, logs, events and documentation to understand an incident.

The problem it solves

Diagnosing a failing pod means crossing several commands and sources, with often cryptic error messages. Existing graphical tools are heavy or do not leverage AI. And sending manifests or logs to an AI service exposes secrets and personal data. Krufter combines fast navigation, assisted diagnosis and privacy by design.

How it works

Connect your clusters: kubeconfig import, multiple contexts, SSH tunnel, one-click metrics server install.
Explore and act: hierarchical navigation, instant filter, command palette, real-time aggregated logs, interactive container shell, pod file explorer.
Ask the AI: on a failing resource the assistant analyses the context, after local masking of sensitive data, and answers with the probable cause and suggested actions.

Features

Fast navigation
Nodes, namespaces, workloads, events; / filter and : palette from the keyboard.

AI diagnosis
Error explanations, probable cause, proposed fixes.

Logs and shell
Real-time aggregated logs, interactive exec, pod file explorer.

Multi-cluster
Several contexts, SSH tunnel, one-click metrics server.

Smart Scrubbing
Local masking of PII, secrets and tokens before any AI call.

Cross-platform
Linux, macOS and Windows, same experience.

AI and privacy

Sensitive data (personal information, secrets, tokens, connection strings) is masked locally on your machine before anything is sent to an AI provider. Your cluster credentials (kubeconfig) are not stored on our servers.

AI requests go through an authenticated proxy that calls Gemini on Vertex AI (Google Cloud, europe-west1 region). A per-plan usage budget protects the service against abuse.

Technology

Core: Rust, kube-rs and tokio, in a hexagonal architecture (pure domain, ports and adapters).
Interface: Flutter (BLoC, feature-first architecture), linked to the core through flutter_rust_bridge.
AI: Google Gemini through Vertex AI.
Cloud platform: Google Cloud / Firebase: Hosting (website), Authentication, Cloud Firestore, Cloud Functions (europe-west1), Vertex AI. Stripe billing; Resend transactional email.
Quality: test-driven development (unit tests, BDD scenarios, Firestore security rules tested on the emulator) and continuous integration building the Linux, Windows and macOS packages.

Product status

Krufter is in private beta, by invitation (version 1.0.0-beta). Sign-up is through the waitlist on krufter.dev; approved testers receive a personal account-creation link and get the Solo plan for free during the beta, then move back to the Free licence.

Plans

Free: discover Krufter (unlimited clusters, 2 devices, fast-model AI diagnostics, core Kubernetes operations).
Solo: personal use (full AI assistant, 200k AI tokens per month).
Pro: power users (SSH bastion tunnel, node root shell, 5 devices, 1M AI tokens per month).
Enterprise: on quote.

Up-to-date plan details are on the home page.

Company and contact

Krufter is edited by the Krufter team. Contact: hello@krufter.com. The website is hosted by Firebase (Google Ireland Limited).

Legal notice · Privacy · Terms