k8sgpt is an Apache-2.0 tool written in Go that scans Kubernetes clusters, diagnoses and triages issues in plain English, built for SREs, platform engineers and DevOps teams who need a fast, readable explanation of why a cluster is misbehaving.

What it is

k8sgpt lives in the Kubernetes and cloud-native ecosystem. It ships as a command-line binary and as an in-cluster operator (k8sgpt-operator), and it inspects the objects, events and state of a running cluster, then turns what it finds into a short, human-readable diagnosis. Its analyzers carry codified SRE experience, meaning the knowledge of what to look for when a workload fails is packaged into the tool rather than left to the operator's memory. Findings can then be enriched by a large language model, with out-of-the-box integration for OpenAI, Azure, Cohere, Amazon Bedrock, Google Gemini and local models.

The concrete problem it solves is manual triage. Instead of reading through cluster events, status fields and logs by hand to work out why a pod, deployment or service is unhealthy, the user runs k8sgpt and gets the relevant signals pulled out and explained in simple English. That replaces the slow, repetitive first pass of incident investigation that otherwise consumes SRE time on every failing workload, and it lowers the level of Kubernetes internals knowledge required to reach a useful answer.

Key capabilities

• Scans Kubernetes clusters and reports diagnoses and triage output in simple English.
• Analyzers codify SRE experience and extract the most relevant cluster information for enrichment with AI.
• Pluggable LLM backends: OpenAI (the default provider), Azure, Cohere, Amazon Bedrock, Google Gemini and local models.
• k8sgpt-operator runs inside a cluster for continuous monitoring rather than one-off CLI runs.
• Operator deployment integrates with existing monitoring such as Prometheus and Alertmanager.
• Model Context Protocol (MCP) support, alongside the documented Analyzers, LLM AI Backends and Key Features sections.
• Distributed as Homebrew formula, RPM, DEB and APK packages, plus Windows binaries, with source under Apache-2.0 and documentation at docs.k8sgpt.ai.

Who uses it and how

• SREs on call use the CLI as a first triage step, letting analyzers surface the relevant cluster facts before digging into manifests.
• Platform teams deploy k8sgpt-operator for continuous cluster monitoring, routing findings through Prometheus and Alertmanager into existing alert pipelines.
• Teams with data-residency or privacy constraints point it at local models instead of a hosted provider, keeping diagnostics inside their own environment.
• Individual engineers install via brew install k8sgpt on Linux or macOS, or use the RPM, DEB, APK or Windows builds on workstations without cluster-wide rollout.

Getting started

Install with brew install k8sgpt on Linux or macOS, or use the published RPM, DEB, APK or Windows binary from the releases page. The default AI provider is OpenAI, so an API key must be generated and configured before AI enrichment works; for in-cluster use, install k8sgpt-operator instead.

How it compares

The provided facts name no paid products that k8sgpt replaces and no directly comparable tools apart from the sister project sympozium, which manages agents in Kubernetes rather than diagnosing cluster issues. On the evidence available, k8sgpt stands alone in this registry.

When to use it — and when not to

A self-hoster must supply an LLM backend: credentials for a hosted provider or a local model, since the default provider is OpenAI and enrichment depends on it. Teams that cannot send cluster diagnostics to an external model service, or that have no appetite for running local inference, should look elsewhere or accept that only non-AI analyzers will be useful. The README excerpt also stops mid-instruction on API key generation, so anyone planning a deployment should read the full documentation at docs.k8sgpt.ai before relying on the repository front page alone.

brew install k8sgpt

brew tap k8sgpt-ai/k8sgpt
brew install k8sgpt

sudo rpm -ivh https://github.com/k8sgpt-ai/k8sgpt/releases/download/v0.4.39/k8sgpt_386.rpm

sudo rpm -ivh https://github.com/k8sgpt-ai/k8sgpt/releases/download/v0.4.39/k8sgpt_amd64.rpm

curl -LO https://github.com/k8sgpt-ai/k8sgpt/releases/download/v0.4.39/k8sgpt_386.deb
sudo dpkg -i k8sgpt_386.deb

curl -LO https://github.com/k8sgpt-ai/k8sgpt/releases/download/v0.4.39/k8sgpt_amd64.deb
sudo dpkg -i k8sgpt_amd64.deb

wget https://github.com/k8sgpt-ai/k8sgpt/releases/download/v0.4.39/k8sgpt_386.apk
apk add --allow-untrusted k8sgpt_386.apk

wget https://github.com/k8sgpt-ai/k8sgpt/releases/download/v0.4.39/k8sgpt_amd64.apk
apk add --allow-untrusted k8sgpt_amd64.apk

==> Installing k8sgpt from k8sgpt-ai/k8sgpt Error: The following formula cannot be installed from a bottle and must be
built from the source. k8sgpt Install Clang or run brew install gcc.

   sudo apt-get update
   sudo apt-get install build-essential

{
  "mcpServers": {
    "k8sgpt": {
      "command": "k8sgpt",
      "args": [
        "serve",
        "--mcp"
      ]
    }
  }
}