# KubeGraf

> Autonomous AI SRE platform for Kubernetes that detects incidents, performs evidence-backed root cause analysis, and delivers dry-run validated SafeFix™ remediations — all running locally without cloud dependency.

KubeGraf is an autonomous, always-on AI SRE platform built exclusively for Kubernetes. It detects incidents like CrashLoopBackOff, OOMKilled, and probe failures, correlates multi-source signals (logs, metrics, traces, events), and delivers evidence-backed SafeFix™ remediations with dry-run validation and human-in-the-loop approval. Powered by OrkasAI, KubeGraf runs entirely local-first — your cluster data never leaves your environment.

- **SafeFix™ Remediation** — *Generates YAML diff previews, blast radius analysis, confidence scores, and one-command rollback for every recommended fix before you apply anything.*
- **Evidence-Based Root Cause Analysis** — *Correlates logs, Kubernetes events, metrics, traces, and recent deployments into a reproducible evidence chain with confidence scores — not a black box.*
- **Dry-Run Validation** — *Simulates every fix using kubectl diff integration before execution; shows exact changes and potential side effects with zero risk.*
- **Anomaly Fingerprinting** — *Detects recurring failure patterns and builds fingerprints to auto-recognize similar incidents, cutting diagnosis time on repeat failures.*
- **Multi-Cluster Management** — *Investigate and remediate incidents across multiple clusters from a single interface without losing investigation context.*
- **BYOK AI Engine** — *Bring your own API key from OpenAI, Anthropic, Gemini, or Ollama; AI calls go directly from your machine to your provider — KubeGraf never sees your key or queries.*
- **Terminal UI + Web Dashboard** — *Use the keyboard-driven TUI during live incidents and the browser-based web dashboard for post-mortems and trend analysis.*
- **Knowledge Bank** — *Local SQLite database stores all incident history; search by pod, namespace, error type, or fix; export reports for post-mortems.*
- **RBAC-Aware Operations** — *Respects your cluster's RBAC policies; suggested fixes adapt to what your user can actually apply.*
- **Full Audit Trail** — *Every analysis, recommendation, and applied fix is logged with timestamps and user context for compliance and post-mortems.*
- **GitOps Integration** — *Sync fixes to Git via ArgoCD or Flux; supports Helm, Istio, Cilium, Nginx, and all major cloud Kubernetes providers (EKS, GKE, AKS, OpenShift, K3s).*

## Features
- Autonomous incident detection (CrashLoopBackOff, OOMKilled, ImagePullBackOff, probe failures)
- SafeFix™ dry-run validated remediations with YAML diff preview
- Evidence-based root cause analysis with confidence scores
- Multi-source signal correlation (logs, metrics, traces, events)
- Anomaly fingerprinting for recurring failure patterns
- BYOK AI engine (OpenAI, Anthropic, Gemini, Ollama)
- Terminal UI and web dashboard
- Local-first architecture — zero data exfiltration
- Knowledge Bank with SQLite incident history
- Multi-cluster management
- RBAC-aware operations
- Full audit trail
- GitOps integration (ArgoCD, Flux)
- One-command rollback
- Human-in-the-loop approval for all changes

## Integrations
AWS EKS, Google GKE, Azure AKS, Rancher, OpenShift, K3s, Helm, ArgoCD, Flux, Istio, Cilium, Nginx, Prometheus, OpenTelemetry, Grafana, OpenAI, Anthropic, Gemini, Ollama

## Platforms
WINDOWS, MACOS, LINUX, WEB, API, CLI

## Pricing
Freemium — Free tier available with paid upgrades

## Version
v1.0.0

## Links
- Website: https://kubegraf.io
- Documentation: https://kubegraf.io/docs/
- Repository: https://github.com/kubegraf/kubegraf
- EveryDev.ai: https://www.everydev.ai/tools/kubegraf