holmesgpt

We break Kubernetes in known ways and grade how an AI incident agent (Aurora) investigates. Every run, including the misses.

AI SRE incident investigation for Kubernetes: HolmesGPT, read-only diagnostics, and reproducible incident cases.

基于 HolmesGPT 的 ERPNext 只读售后诊断工作台：Vue 3 / TypeScript、FastAPI、PostgreSQL、Celery、Valkey、多账号授权、证据与审计、Docker 隔离部署。

Testuję narzędzia DevOps i AI dla DevOps na prawdziwych awariach. Pełne logi z każdego przebiegu.

Kubernetes-native AI SRE: alert investigation with HolmesGPT, guarded autofix, multi-cluster and host agents, pluggable LLMs

Demo: kubectl-guard as a guardrail + audit layer around an agentic Kubernetes workflow (HolmesGPT), with Prometheus/Grafana observability.

Enterprise OpenTelemetry on EKS: OTel collector pipelines, 2-tier gateway, eBPF, LGTM (Loki, Grafana, Tempo, Mimir), Google SRE SLO alerts, 70–90% FinOps savings & AI-ready runtime (k8sgpt, HolmesGPT) for SRE agents.

Teleport Machine & Workload Identity demo: an AI SRE (HolmesGPT) investigates Kubernetes with its own tbot identity — bound-keypair join, short-lived certs, RBAC-scoped, every call audited