Repolode
Submit
sno-ai

sno-ai/sno-station

Lets Codex, Claude Code, Cursor, Hermes Agent and OpenClaw share one encrypted local memory, hand off tasks, and review each other's work.

View on GitHubWebsite
Stars
765
Forks
363
Open issues
4
Last commit
yesterday

What is it?

Sno Station connects the AI coding agents you already use so they work as one team. Codex, Claude Code, Cursor, Hermes Agent and OpenClaw read from the same memory, so when one hits a rate limit, another can pick up the task with its context intact. Any agent can also ask a different agent to review its work, so the reviewer does not simply agree with the first one.

The memory, messages and skills are stored on your own computer, and the memory is encrypted with a key that stays on that machine. No account or server is required to run it. A nightly job called REM consolidates memory across sessions and runs Self Learning, which reads the day's sessions, finds repeated mistakes and proposes changes to the agents' skills. You review each proposal and nothing changes until you accept it.

An optional sno.ai account adds a dashboard for reviews, learned lessons and skill changes. Cloud features such as REM Enhanced mode send memory text to Sno's servers, while Local First mode keeps processing on your machine. Cursor support is limited to the cursor-agent command line on Linux and macOS and the Cursor app on macOS; Windows is not supported.

Key features

  • 01Shares one encrypted memory across Codex, Claude Code, Cursor, Hermes Agent and OpenClaw.
  • 02Hands unfinished work to another agent when one runs out of quota.
  • 03Lets one agent request a review from a different agent.
  • 04Runs a nightly job that proposes skill changes from repeated mistakes, applied only after you accept.
  • 05Keeps memory and skills on your computer, with optional cloud features.

Use cases

  • Continuing a long coding task in another agent after one hits its usage limit.
  • Getting a second agent to review code written by the first.
  • Keeping preferences and project context consistent across several AI coding tools.
  • Reviewing overnight suggestions for improving your agents' skills.

Topics

#ai-agents#coding-agents#agent-memory#multi-agent#code-review#local-first#claude-code#developer-tools#agent-skills#codex

FAQ

What is sno-station?
Lets Codex, Claude Code, Cursor, Hermes Agent and OpenClaw share one encrypted local memory, hand off tasks, and review each other's work. Sno Station connects the AI coding agents you already use so they work as one team. Codex, Claude Code, Cursor, Hermes Agent and OpenClaw read from the same memory, so when one hits a rate limit, another can pick up the task with its context intact. Any agent can also ask a different agent to review its work, so the reviewer does not simply agree with the first one.
Is sno-station free to use? What license does it have?
sno-station is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is sno-station still maintained?
Actively maintained. The last commit to sno-station was on Oct 21, 2026.
What can I use sno-station for?
Continuing a long coding task in another agent after one hits its usage limit. Getting a second agent to review code written by the first. Keeping preferences and project context consistent across several AI coding tools. Reviewing overnight suggestions for improving your agents' skills.
Where can I get sno-station?
The source code and installation instructions are on GitHub: https://github.com/sno-ai/sno-station

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects