probing api…
managed hermes agent hosting

Hermes Agent hosting,
live in two minutes.

Cagebox is managed hosting for Hermes Agent. Connect Telegram or Discord, paste your model key, and your agent answers around the clock — from its own hardware-isolated microVM. No Docker, no SSH, no Caddy snippet to copy. You bring the keys and the messages; we bring the VM, the subdomain, and the webhook plumbing.

~/agents/hermes-concierge● running
$ hermes status
runtime    Hermes Agent
channel    Telegram + Discord
memory     persistent /data
isolation  Firecracker microVM
idle       snapshot · ~590ms wake
Isolated
Hardware microVM, private kernel
Always-on
Answers around the clock
Persistent
Memory across conversations
Your keys
OpenAI / Anthropic / OpenRouter

Hardware-isolated, not a container

Every Hermes agent runs in its own Firecracker microVM with a private kernel — the same hypervisor AWS Lambda uses for untrusted code. Your prompts, memory, and API keys sit behind a CPU boundary, not a Docker namespace. See the security model.

Operate it from one dashboard

Live logs, a real web shell, a file browser for /data, and a resource monitor — all in the browser. Inspect what your agent is doing, edit a config, or restart it without SSH or a separate host.

Persistent memory and a browser tool

Hermes keeps memory across conversations on persistent /data and can browse the web when it needs to. Perfect for a private concierge that remembers you, or a help-channel bot that summarises long threads on demand.

Telegram & Discord, wired for you

Paste a Telegram bot token or sign in to Discord and Cagebox registers the webhook against your private *.cagebox.dev subdomain automatically. No tunnels, no public IP to manage.

Bring your own model key

OpenAI, Anthropic, OpenRouter, Groq, Mistral, and more — pick the provider, paste the key. Stored encrypted and decrypted only inside your VM; model usage is billed directly by the provider.

From signup to a live Hermes bot

  1. Pick the Hermes runtime — one radio button.
  2. Connect a channel — Telegram token or Discord.
  3. Bring your model key — pick, paste, done.
  4. Hit create. ~45s to your first reply.
FAQ

Hermes Agent hosting, answered.

What is managed Hermes Agent hosting?

Cagebox runs your Hermes Agent for you in its own Firecracker microVM. You connect Telegram or Discord and paste a model key; we handle the VM, subdomain, webhooks, snapshots, and updates. No Docker, SSH, or server to maintain.

Do I bring my own model API key?

Yes. Bring an OpenAI, Anthropic, OpenRouter, Groq, Mistral, or other provider key. Keys are stored encrypted and decrypted only inside your VM — model usage is billed directly by the provider.

Is my Hermes agent always reachable?

Yes. Your agent stays reachable around the clock. Between messages it is snapshotted to disk and resumes in under a second when someone pings it — so it always answers without you running or babysitting a server.

Host your Hermes agent today.

A live agent in two minutes, hardware-isolated, answering around the clock. Bring your model key and ship.

Request invite