s13n

Every agent knows what the others know.

Your team settles something in one chat, and the next agent starts from zero. s13n keeps decisions, conventions and memory in sync across Claude, Cursor, ChatGPT and the agents you build yourself.

Private beta. One email when your seat is ready, nothing else.

ctx://acme/checkout 4 agents
Decision from claude-code, just now

Prices are formatted on the server. The browser never does money math.

  • claude-code wrote the decision Synced
  • cursor editing cart.tsx Synced
  • chatgpt drafting release notes Synced
  • support-bot has an older rule Conflict
1 conflict waits for you. s13n never overwrites quietly.
Illustration

How it works

You connect once. After that, writing something down is the only step you repeat.

  1. 1

    Connect your agents

    Add s13n as an MCP server in Claude Code, Cursor, ChatGPT or your own agent. No SDK to learn.

  2. 2

    Write it down once

    A decision, a naming rule, a client's preference. Any agent or person can add it, and s13n keeps who wrote it and when.

  3. 3

    Every agent starts from it

    Before an agent answers, it reads the context its project allows. When two sources disagree, you decide.

What you get

Speaks MCP

Works with any agent that supports the Model Context Protocol. Nothing in your agents to rewrite.

claude-code, cursor, chatgpt

Conflicts surface

When two sources disagree, you see both side by side and keep one. Nothing gets overwritten without you.

Conflict

Scoped to the work

Context lives per project and per team. An agent sees what its task needs and nothing more.

ctx://acme/checkout

Who wrote what

Every entry keeps its author and time, agent or person, so any rule traces back to where it came from.

claude-code, 14:02

Plain files out

Export everything as Markdown, any time. Your context stays yours.

context.md

Questions

Why the name?

s13n is short for synchronization: an s, thirteen letters, an n. The same way i18n stands for internationalization and k8s for Kubernetes.

Can I use it today?

Not yet. We are building it with a small group of teams first. Join the waitlist and you get one email when there is a seat for you.

Which agents will it work with?

Anything that speaks MCP. Today that covers Claude, Cursor, ChatGPT and most agent frameworks, including the ones you write yourself.

Where does my context live?

In your team's workspace, and you can export all of it as Markdown whenever you like. We will publish the full data policy before the beta opens.

Stop explaining your project to every agent from scratch.

Join the waitlist