Sharing a Namespace in Skein: What It Is and How It Stays Yours
There is a moment that shows up once you have been using AI tools on real work. You have built up context with your assistants about a client, a project, or something in your personal life. Then someone else joins, a teammate, a subcontractor, the client themselves, and they start from nothing. They ask their own AI the same questions, get vaguer answers, and you end up re-explaining the background you already taught your tools weeks ago.
Sharing a namespace is the fix. You hand one slice of your memory to someone, and their AI tools read the same thread yours do. Here is what that means, why it helps, and how it stays under your control.
What a shared namespace is
A namespace is one space in your memory. You might keep work in one, personal life in another, and a specific client or project in its own, so the right context shows up in the right place and nothing bleeds between them.
Sharing takes one of those spaces and lets another person's AI tools read it. Not your whole memory. One space, the one you choose. Everything else stays private to you.
Why you would want it
The pattern is always the same: more than one person needs the same background, and right now each of them is teaching it to their own tools separately.
- A consultant keeps a client's project in its own space. The client's team connects, and everyone's assistant answers from the same context instead of each person rebuilding it.
- You share a space for a family trip you are organizing, so the people coming along can have their assistants check the plan without asking you for it.
- You hand a space to a collaborator for the length of a project, and take it back when the work is done.
You teach the context once, to the space, instead of to every person's tools one at a time.
How it stays yours
This is the part that makes sharing something you can do without second-guessing it.
It is read only. The people you share with can read the space. They cannot change it, add to it, or delete anything. What they see is what you put there.
It is opt-in on both sides. Before anyone can share with you, the two of you connect, and connecting takes a yes from each of you. After that, a space someone offers only appears for you once you accept it. Nobody lands in your memory by surprise, and you never land in theirs.
They see the space and nothing else. Someone you share a namespace with sees your public profile and the thoughts in that one space. They do not see your other spaces, your other connections, or anything else about your account.
You can see who wrote what. Every memory in the space still records which tool wrote it, so a reader can tell what came from you and what came from one of your assistants. Shared does not mean anonymous.
You can take it back. Share a space with as many people as you want, and remove anyone's access at any time. Removing one person leaves the space untouched for everyone else.
Your own agents are a separate control
Sharing with other people is read only on purpose. Your own agents are different. You decide which of your agents can reach each namespace, and whether each one can read it, write to it, or have full access. So an agent that works on one project gets exactly that space, at the level you set, and nothing outside it. That keeps an autonomous tool useful without handing it the run of your whole memory.
How this is different from other memory
Most AI memory is built for one person. The memory inside ChatGPT or Claude stays on your account and cannot be handed to anyone else. A notes app can share a document, but that is a file you both open, not a memory your tools read on their own. A shared namespace is the second thing: a slice of your memory that someone else's tools read through MCP, while everything else stays yours.
For how tools connect at all, see What Is MCP?. For how each connection stays scoped and revocable, see OAuth vs API Keys and How Do You Trust an AI Memory?.
Who it is for
Sharing is on the Pro and Max plans. It is for people whose work already spans other people: clients, teammates, a collaborator on a project, or family on something you are organizing together. If your AI work is only yours, you may never need it, and the rest of Skein works the same without it. The day a second person needs the context you have already built is the day it earns its place.