Homelab tinkerer in upstate NY. I build small, focused tools for self-hosted media stacks — Claude Code skills, WebDAV/Usenet streaming, and the occasional editor plugin. Day-to-day: keeping a real Arr-stack healthy and learning out loud.
Operating a production Arr-stack day-to-day — Radarr, Sonarr, Lidarr, Readarr, Prowlarr, plus a download client and request manager. Most public work comes from scraping that experience into reusable tools.
I lean on Claude Code as a homelab operator's toolbelt — configuration sync, TRaSH-Guides profile application, health sweeps, and other things I'd rather not do by hand at 2am.
Contributing to the nzbdav super-fork: WebDAV-mounted NZBs, on-demand streaming, and a SABnzbd-compatible API so Sonarr/Radarr treat it as a drop-in download client.
Claude Code skills for running a self-hosted Arr-stack (Radarr/Sonarr/Lidarr/Readarr/Prowlarr) — config backup & diff, TRaSH-Guides quality profiles, request-manager wiring, download-client queue ops, path validation, secrets handling, health sweeps, and compose lifecycle. Talks to each app's real REST API; no bespoke dashboard, no hardcoded infra.
A maintained super-fork of nzbdav-dev/nzbdav. Mounts
NZBs as a virtual filesystem over WebDAV and streams straight from
Usenet with full seeking — no full download required. Also
exposes a SABnzbd-compatible API so Sonarr/Radarr can use it as
a drop-in download client. Early adopters report ~2× network
throughput and ~4× faster seek times vs. v0.6.4.
Small gedit plugin that adds a status-bar label with a
live word count of the active document. Word is defined as
[a-zA-Z0-9]+[a-zA-Z0-9\-']*. Kept around for
historical reasons — I still use it on the rare day I open gedit.