Note/To access pods using terminal commands, apply for our early beta test role in our Discord

pods.ml
Open dashboard

Spin up an
entire agent stack in
thirty seconds.

Pick a template, hit deploy, get a public URL. Reproducible runtime, your data, one box.

Deploy your first podSee how it works
how it works3 steps

From template to public URL in three steps.

  1. 01Pick a template

    Hermes agent, code sandbox, n8n, Minecraft. Each one is a vetted pod with its image, resources, ports, and post-deploy surface defined ahead of time.

    templates
    4
    families
    agent, automation, sandbox, game
    form
    rendered per template
    $pods list
  2. 02Hit deploy

    Your pod is provisioned in an isolated container with its own filesystem, given a port, and started. The deploy view streams live install progress back to your browser.

    runtime
    isolated container
    public url
    per-pod subdomain
    tls
    automatic
    $pods deploy hermes --auto-domain
  3. 03Ship the URL

    Inside thirty seconds you have a public HTTPS endpoint. Open the in-app terminal, watch live CPU and memory, attach connectors, and swap providers without leaving the dashboard.

    latency
    p50 around 27s
    surface
    https, tcp
    terminal
    live in browser
    $open https://juno.pods.ml
the pod family4 templates

Four pods today. More coming.

  1. 01

    Hermes Agent

    agent

    Always-on AI agent with persistent memory, skills, MCP servers, and over 30 messaging connectors. The flagship pod, the reason this exists.

    port
    8080
    memory
    4 GB
    proto
    https
    cpu
    1 vCPU
    Deploy
    agentlive

    Hermes Agent

    $ pods deploy hermes

    Always-on AI agent with persistent memory, skills, MCP servers, and over 30 messaging connectors. The flagship pod, the reason this exists.

    port
    8080
    proto
    https
    memory
    4 GB
    cpu
    1 vCPU
    disk
    15 GB

    Skills, MCP servers, connectors for WhatsApp, Discord, Slack, and Telegram. OAuth handoff, streaming terminal, and a managed mailbox per pod.

    Deploy Hermes Agent
  2. 02

    Code Sandbox

    sandbox

    Ubuntu pod with VS Code in the browser, Claude Code CLI, or a plain shell. Persistent home directory, sudo, and a public URL.

    port
    8080
    memory
    4 GB
    proto
    https
    cpu
    1 vCPU
    Deploy
    sandboxlive

    Code Sandbox

    $ pods deploy code-sandbox

    Ubuntu pod with VS Code in the browser, Claude Code CLI, or a plain shell. Persistent home directory, sudo, and a public URL.

    port
    8080
    proto
    https
    memory
    4 GB
    cpu
    1 vCPU
    disk
    20 GB

    Persistent home directory, sudo access, and a public HTTPS URL out of the box. Bring your own toolchain or pick a preinstalled flavor.

    Deploy Code Sandbox
  3. 03

    n8n

    automation

    Visual workflow automation with drag-and-drop nodes, 400 plus integrations, and queue mode. Public webhook URL out of the box.

    port
    5678
    memory
    2 GB
    proto
    https
    cpu
    1 vCPU
    Deploy
    automationlive

    n8n

    $ pods deploy n8n

    Visual workflow automation with drag-and-drop nodes, 400 plus integrations, and queue mode. Public webhook URL out of the box.

    port
    5678
    proto
    https
    memory
    2 GB
    cpu
    1 vCPU
    disk
    8 GB

    Queue mode, 400 plus integrations, and a public webhook URL out of the box. Editor protected by basic auth.

    Deploy n8n
  4. 04

    Minecraft (Paper)

    game

    High-performance Bukkit and Spigot compatible server. Latest stable Paper build, console plus file access plus plugin folder.

    port
    25565
    memory
    4 GB
    proto
    tcp
    cpu
    2 vCPU
    Deploy
    gamelive

    Minecraft (Paper)

    $ pods deploy minecraft-paper

    High-performance Bukkit and Spigot compatible server. Latest stable Paper build, console plus file access plus plugin folder.

    port
    25565
    proto
    tcp
    memory
    4 GB
    cpu
    2 vCPU
    disk
    10 GB

    Console plus file access plus plugin folder. Latest stable Paper build, server.properties editor, and version pinning.

    Deploy Minecraft (Paper)
self-hosting3 reasons

The data is yours. The runtime is yours. The bill is yours.

“If your agent platform can read your prompts, it’s not your agent platform.”
The pods.ml thesisnorth star, always
  • your domainhttps://*.pods.ml
  • edgeautomatic tls, per-pod subdomain
  • dashboardauth, deploy, live console
  • your podisolated runtime, your data
  • the modelyour provider, your key
  1. 01principle

    Self-hosted from byte one.

    The whole platform runs inside your own Azure, Hetzner, or bare-metal box. We never see your traffic, your prompts, or your model keys.

  2. 02principle

    No middleman cloud.

    Pod data lives on your disk. No managed database, no telemetry pipe, no third-party inference relay sitting between you and the model.

  3. 03principle

    One box, many pods.

    A single 4-vCPU box runs an agent, an automation server, a code IDE, and a game server side by side. Pay your cloud bill, not ours. Resources are hard-capped per pod, not by us.

№ 002, the close

One click.
One pod.
Ready.

Sign in, pick a template, and you have a running agent with a public URL before your coffee cools. Free during the prototype.

Open the dashboardRead the source