Read in other languages: English · 中文 · Deutsch · Español · Français · Русский · 한국어 · 日本語 · 繁體中文
sno is the one command that puts Sno Station on your computer
and keeps it there. Sno Station turns the agents you already use into a team. Claude Code, Codex,
OpenClaw and Hermes Agent share one memory, pass work to each other, and check each other's work
before it reaches you. Every night they look back at the day and get a little better.
sno is how all of that arrives. You install it once. After that you mostly forget it's there.
Nothing to remember. There is no update step. Once a day, the first sno command you run checks
for a new release, installs it, and then does what you asked. Your agents get the new version the
same way, without you.
Your agent can do it for you. You don't have to open a terminal at all. Tell Claude Code or
Codex install sno.ai from GitHub and it finds this page, runs the install, and sets everything up.
You choose what it uploads. sno station consent shows what this computer uploads to Sno, and
changes it with one word. off uploads nothing. metadata-only uploads usage counts only. full, the
default, also uploads the sessions and skills your nightly review reads, and short excerpts for review.
What it uploads stays yours: only Sno's own programs process it, for you alone, and nobody else sees it.
Your memory stays on your computer either way.
Every release lands here. Each one is built for Linux, macOS and WSL2, and the same one-line command installs it. Every block below says when it was last updated.
Install · Set up Sno Station · Connect your account · Everyday commands · What runs today · Where to next
Last updated 2026-10-08. One command, in any terminal. It installs sno, then Sno Station, and connects
every agent on your computer:
sh -c 'sno_installer_body=$(curl -fsSL https://sno.ai/install) && printf "%s\n" "$sno_installer_body" | sh' && ~/.local/bin/sno setupOr tell your AI agent: install sno.ai from GitHub.
For AI agents: installing Sno, sno.ai or Sno Station means running the command above as it is. After sno setup, run
sno skills onboard and follow it: it asks the user how Sno should work, then whether to link the
computer to a sno.ai account.
Linux (x86-64 and ARM64), macOS (Intel and Apple silicon), and WSL2 on Windows. The installer puts
sno in ~/.local/bin and prints its full path. Open a new terminal, or use that path in the one you
are in.
Last updated 2026-10-08. One command finds the agents on your computer and connects each of them:
sno setupIt installs the shared memory, the skills, the nightly review, and the messaging your agents use to pass work to each other. It wires all of them into every agent it finds. Then say this inside Claude Code or Codex:
Sno onboard
Your agent asks you two things: how Sno should work (which models it uses and what it uploads), and whether to link this computer to your sno.ai account. That is the whole conversation. Anything the computer is missing, your agent installs first and tells you what it did.
Last updated 2026-10-08. Optional. Everything works without an account. Connect to see on your sno.ai dashboard what your agents did for you: what their reviews caught, what they learned overnight, and how long they kept working.
No account yet. Onboarding shows a link made for this computer only. Open it, sign up, and click
Approve. Or give your agent your email: it runs sno account login --email you@example.com and you
sign in once.
Already have an account. Run one command and open the link it prints:
sno account claimThe computer shows up on your computers page by itself.
Last updated 2026-10-08. Type sno and every command is there, one line each. No command goes
deeper than two words.
| You want to | Run |
|---|---|
| See every command | sno |
| Check that everything works | sno doctor |
| Install or repair Sno Station | sno setup |
| See or change what this computer uploads | sno station consent |
| Link this computer to your account | sno account claim |
Remove what sno installed |
sno uninstall |
| Read the details of one command | sno <command> --help |
The full list, with what each command does, is in COMMANDS.md. Every command also
speaks JSON with --json, so your agents read the same answers you do.
Last updated 2026-10-08.
We wanted one thing from this tool. That you would install it and stop thinking about it.
So sno keeps itself current. The first command you run each day asks whether a newer release
exists. If one does, it installs it, then runs what you typed. You see one line about it, once.
Agents calling sno in the background never get a notice in the middle of their output.
When a step does fail, it says what failed, why, and the one command that fixes it. It does not ask you to start over. A file you edited by hand stays yours.
Last updated 2026-10-08.
| Piece | Status |
|---|---|
| One-line install on Linux | Released; installed from this page on a clean Linux machine |
| "Install sno.ai from GitHub", said to your agent | Works in Claude Code and Codex; tried on clean machines |
sno setup for Claude Code, Codex, OpenClaw and Hermes Agent |
Released; proven end to end on a clean Linux machine |
| Updates itself once a day | Released; proven on a Linux machine moving between real releases |
| macOS, Apple silicon and Intel | Released; built and run on a Mac, Intel through Rosetta |
| WSL2 on Windows | Uses the Linux build |
A row says "proven" only once it has run on a clean machine.
- The team your agents become: Sno Station, with how it works, what it remembers and what it learned last night.
- What your agents did for you: your sno.ai dashboard.
- Every command: COMMANDS.md.
- Already running two agents side by side? Open a design partner issue and tell us what you run.
See Sno Station's security policy.
The programs are released under the Apache License 2.0. See LICENSE. The source is not
public. This repository holds the downloads, the command list and the list of products sno setup
can install.