Blog
3posts. Back to all posts.
How I am building an alert-handling layer on top of Wazuh: a local LLM writes first-pass summaries, I triage with a written playbook, and a council of models gates any automated patching, failing safe to high risk.
Read Private repository
Why I wrote my own Model Context Protocol servers for Proxmox, Wazuh, Tailscale and Jira, and the scoped, read-mostly design that keeps an AI agent useful without handing it the keys.
Read Private repository
A home security monitoring stack built around two complementary lenses: Wazuh for hosts and as the central log and alert pipeline, Suricata for network traffic, and what I learned about tuning them.
Read Private repository