Hermes Agent VPS hosting is now live on LumaDock

Short version: Hermes Agent is now a 1-click template on every LumaDock VPS plan. You pick a plan, the server boots with the agent pre-installed on Ubuntu 24.04, the dashboard ready to pair over SSH. The "from order to first answered message" loop is under five minutes on a normal connection.

For people who already know Hermes Agent that's the whole story. Skip down to the bottom for the links. The rest of this post is for the larger group of self-hosters who keep seeing Hermes in their feed and want a quick honest read on what it actually is and why it's worth our 1-click slot.

What Hermes Agent is

Hermes Agent is the open-source self-hosted AI agent that Nous Research shipped in February 2026. MIT-licensed end to end, no telemetry, runs on a $5-class VPS. Think of it as the bit that sits between an LLM provider (Anthropic, OpenAI, Nous Portal, OpenRouter, anything with a vaguely OpenAI-compatible API) and your messaging channels (Telegram, Discord, WhatsApp, Slack, Signal, Email and a long tail of others). It remembers what you tell it across sessions, learns skills from repeated tasks and exposes all of it as flat markdown files plus a SQLite database you can grep.

The thing that makes Hermes feel different from a stock LLM chat is the learning loop. After a few successful runs of similar work, the agent generates a SKILL.md capturing the procedure. Next time you ask for the same kind of task, it's quicker, more confident, less likely to wander. The skill files live in your data directory, so you can read them, edit them, delete the ones that learned the wrong lesson. The whole memory model is debuggable in a way that "trust the model" agents aren't.

If you've been running OpenClaw and reading the security writeups from March, Hermes is the project most people in that crowd have been moving to. The migration tool ships with the agent and handles persona, memories, skills and API keys in one go.

Why we built the template

Two reasons.

The first is volume. Hermes hit a hundred thousand GitHub stars in under two months. We watched the support tickets in our OpenClaw template tilt heavily toward "how do I switch to Hermes" through April. When that kind of demand shows up, the next move is to make the destination as boring to deploy as the source.

The second is friction. The manual install is a 90-second curl-pipe-bash command, which is fine for people who already live in a terminal. For everyone else, there's still a setup wizard to walk through, providers to pick, env files to edit, systemd to wire up. We wanted the "I just want to chat to an agent over Telegram" use case to be one click instead of an hour. The template gets you there.

What's in the template

You get Ubuntu 24.04 with Hermes Agent pre-installed under the standard install layout. The hermes CLI is on the PATH, the data directory at ~/.hermes is created, the systemd unit is ready and the gateway dashboard is bound to localhost on port 9119. On first SSH login you see a friendly setup menu that points you at either the web dashboard (recommended for most people) or the terminal wizard (if you'd rather not open a browser).

The web dashboard route uses an SSH tunnel so the dashboard never gets exposed to the public internet, even though it can edit API keys and bot tokens. Reach it from your laptop with one command, configure providers and channels in a browser, close the tunnel when you're done. The Hermes Agent VPS setup guide in our knowledge base walks the exact steps and shows you the screens.

The terminal wizard is what you'd run with hermes setup on any Hermes install. It offers a Quick setup that picks a provider, a default model and an optional messaging channel in about five minutes, or a Full setup if you want every option exposed.

Sizing-wise, Hermes runs fine on any 2 GB plan. If you're planning to lean on browser automation or run multiple agents, the 4 GB and 8 GB tiers give you more headroom without a sharp price jump.

Where it makes most sense

Three rough audiences will get the most out of this template.

Developers who want a personal assistant they actually control. The agent is on your VPS, the memory is in flat files you can read, the LLM provider is whoever you tell it to use. No vendor-locked chat history living on someone else's server.

OpenClaw users who've decided to switch. The migration tool brings your persona, memories and skills across in one command. Run it on the new box from a tarball of your old ~/.openclaw and your agent feels like home within a minute. Side-by-side operation works too if you'd rather trial Hermes for a few weeks while leaving OpenClaw running on the old box.

Small teams who want a self-hosted Slack or Discord bot with persistent memory of project context. Hermes handles multi-channel gateways out of the box, so the same agent answers in Slack at work, Telegram on your phone and email overnight, all with one shared brain.

Hermes vs the other 1-click agents we host

If you're new to self-hosted agents and not sure which one to pick, the cheat sheet:

• OpenClaw is still the most mature option. Bigger skill library, longer track record on every messaging platform, slower release cadence. Good fit if you want a tool that's been battle-tested for a year.
• Hermes Agent is the newer architectural pick. Cleaner memory model, learning loop, faster release cadence, smaller attack surface. Good fit if you want what's shipping now.
• ZeroClaw is the lighter option. Less heavy, less learning behaviour, easier to reason about. Good fit for setups where you want a predictable bot rather than an autonomous one.

For a real comparison rather than a marketing pitch, the Hermes Agent vs OpenClaw self-hosting comparison in our tutorials walks the operational trade-offs in detail.

Getting started

Three paths depending on where you are right now.

You want to try Hermes Agent without thinking about install. Pick the Hermes Agent VPS template when ordering. Boot the box, SSH in, follow the menu, configure your provider, send the bot a message. Five minutes.

You're already running OpenClaw and want to switch. Start with the when-to-migrate decision tutorial, then the migration walkthrough. The hermes claw migrate command does the heavy lifting; the tutorials cover the pre-flight checks and the post-migration verification.

You want to install Hermes Agent on a server you already have. Our Ubuntu install tutorial walks the curl-pipe-bash flow plus the manual fallback for when the installer chokes. The template is the gentler route; the install guide is the deeper one.

For everything else (production hardening, backups, messaging gateways, troubleshooting, the memory model, cost optimisation), our tutorials section has a growing library and you can find the curated tour through them in our Hermes Agent complete guide.

If you do try the template and find a rough edge, ping support and we'll fix it. The 30-day refund window means you can take it for a real spin without committing.