News.
From the PodWarden team
Your instance is now reachable from every Hub node
A connection error that could outlast a deploy is gone. Every Hub node can now reach every connected instance — here is what was happening, and what changed.
Catalog volumes now classified, mount by mount
Every template in the catalog now carries a per-volume intent map. Backups, storage classes, and migrations can act per-mount instead of per-stack — 3,801 volumes across 2,168 templates classified into seven buckets.
vLLM Warden is now open source
We've open-sourced vLLM Warden under Apache-2.0 — a browser UI, OpenAI-compatible gateway, and model lifecycle manager that wraps vanilla vLLM. Available on GitHub and one-click installable from the PodWarden Hub catalog.
Automatic image cleanup on all Hub nodes
A weekly systemd timer now prunes the containerd image store on every PodWarden Hub production node. First run: Sunday 2026-05-24 04:00 UTC.
Hermes Agent Is Now in the Catalog
Nous Research's self-improving AI coding agent — with persistent memory, a skills library, and multi-platform messaging — is now a one-click deploy on PodWarden.
Ollama and Open WebUI: Run Local LLMs on PodWarden
Ollama brings local LLM serving to your PodWarden cluster, and Open WebUI wraps it in a ChatGPT-style interface. Together they give you full control over your AI workloads — no API keys, no data leaving your network.
AirPipe: P2P File Transfer, Now in the Catalog
AirPipe is a self-hosted, end-to-end-encrypted file transfer with passphrase pairing and WebRTC peer-to-peer. New featured app in the PodWarden catalog.