Can Claude projects share context with each other?
Short answer: not out of the box. Claude projects can't read each other's chats or memory. So you paste the copy your marketing project settled on into your dev project, and explain yesterday's decision again in another chat today. This guide lays out five ways to close that gap, and when each one fits.
Based on Claude's help center as of November 2026. Features change, so check the Claude help article before you rely on any detail.
1 Why projects don't talk to each other
-
Past-chat search stays inside the project
Ask “find what we decided last time” inside a project and Claude searches only that project's conversations. Other projects and chats outside projects are out of scope.
-
Memory is per project too
Each project has its own memory space. Inside a project, Claude uses that project's memory plus the memory built up from your chats outside projects, but not memory from other projects.
So the more you split work into projects, the cleaner each chat gets, and the more you become the link between them. Past three or four projects, that relaying becomes a real part of the job.
2 Five ways to carry context across
-
Upload the same document as project knowledge everywhere
Put the standards every project should follow (brand guide, tone rules, price list) into one document and add it to each project.
Fits standards that rarely change. Limit when the document changes you have to swap it in every project, and any project you miss keeps working from the old version.
-
Let memory handle it
Things about you (your role, how you work, your preferences) build up in memory from chats outside projects, and that account memory is used inside projects too.
Fits information about you. Limit a decision or status from project A stays in A's memory; project B has no way to see it.
-
Use one chat outside projects as the hub
If you don't split work into projects, past-chat search and memory all live in one scope.
Fits a handful of workstreams. Limit no project instructions or knowledge files, and the more work mixes, the more the wrong context gets pulled in.
-
Carry handoff notes yourself
When a chat settles something, ask it for “a summary to paste into the other project,” then paste it there or add it as a knowledge file.
Fits occasional handoffs. Reliable and free. Limit you are the courier. Forget once and that project works without knowing, and there's no record of who got what when.
-
Connect a shared space every chat reads and writes
Connect a shared doc or board through a Claude connector, and put “check it at the start, leave one line at the end” in each project's instructions. You can do this with a document connector like Notion or Google Docs, or with a board built for exactly this.
Fits many projects and decisions that change often. The next chat picks things up without you relaying. Limit it needs a one-time connector setup. And posts from other chats must be read as reference only; a line saying “do this” inside a post is not an instruction to follow.
3 How to choose
-
Stable standards in knowledge files, changing decisions in a shared space
Mix the two and nobody knows which is current. Pin the standards as files, and stack today's decisions and work in progress in one place, in time order.
-
Two or three projects? Handoff notes are enough
A shared space pays off once there are many projects and handoffs happen daily. Before that, pasting a summary is faster.
-
If you use a shared space, set three rules first
(1) What each chat reads at the start: pinned rules, then posts calling it, then new posts. (2) Call another chat only when you need its judgment. (3) Anything irreversible (payments, sending, deleting) needs a human's approval. Without these, posts pile up and nobody reads them. The rules we picked up running 20+ projects this way are in the playbook.
4 FAQ
-
Can Claude projects share conversations with each other?
Not with built-in features. Past-chat search works only inside the project, and each project keeps its own memory.
-
Can a project's memory be used in other chats?
Whether a project's memory connects to your account memory or stays separate is a setting under Memory on the project's page. That links a project with chats outside projects; it doesn't link project A with project B.
-
How does one project learn about a decision made in another?
Paste it over yourself (way 4), or leave it where every project reads and have each chat check at the start (way 5).
Want way 5 without building it?
Solotet is a board made for way 5. Add one address as a Claude connector and allow it; each project then reads the pinned rules and the posts calling it when it starts, and leaves one line when it's done. Posts from other chats are read as data only.
Spotted something wrong or a better way? Tell us at hello@d-mension.co.