Claude Code Projects shared memory vs Memlo
Last updated
Claude Code Projects, in beta since September 17, 2026, runs a goal as parallel Claude Code cloud sessions, called threads, under a coordinator, and every thread adds to and draws from a shared memory. It is a good fit for Claude-only work that runs in the cloud. Memlo is a memory server that agents from different vendors share, including local Claude Code and Codex sessions, ChatGPT and claude.ai, and that you can share with people, with every version of every memory kept. Use Projects to split one goal across Claude threads; use Memlo when the memory has to outlive one project, cross tools or reach your team.
What Claude Code Projects is
Anthropic describes the redesigned Projects this way, in a post dated September 17, 2026:
- "Projects have threads that do the work and a coordinator that directs them."
- "Each thread is a Claude Code cloud session working on its own branch and copy of the repo."
- "Every thread now adds to and draws from a shared memory, reducing the need for complex prompt engineering."
- "Threads run in the cloud today; running on your machine alongside your local tools and code and behind your network is coming very soon."
The beta is "available in beta to select Claude Pro and Max subscribers who use cloud sessions in Claude Code and don't have any existing projects on the web or desktop." Team and Enterprise plans come later.
At a glance
Facts about Projects come from Anthropic's announcement, read on October 30, 2026. Links are in Sources at the end.
| Claude Code Projects | Memlo | |
|---|---|---|
| What it is | A coordinator and parallel Claude Code cloud threads working toward one goal | A hosted MCP memory server your agents share |
| Splits the work | Yes: the coordinator scopes, delegates and assembles the result | No: you or your lead agent split the work |
| Where agents run | Claude Code cloud sessions today; on your machine "coming very soon" | Wherever your agents run: local terminals, cloud sessions and chat apps |
| Which agents | Claude | Claude Code, Codex, claude.ai, Claude Desktop, ChatGPT, Gemini and other MCP clients that sign in with OAuth |
| Memory scope | Shared by the threads of one project | Spaces you choose, used by any project and any connected agent |
| See and edit the memory | The announcement doesn't describe it | Each memory is a titled note you can read and edit in the Memlo app or through an agent |
| History | The announcement doesn't describe it | Every change is a version, with when it was written and by whom; any version can be restored |
| Share with people | The announcement doesn't mention it | Share a space or a page by email as a Viewer or an Editor, or with a team |
| Who can use it | Select Pro and Max subscribers, in beta | Invitation-only beta; join the waitlist |
| Price | No separate price in the announcement | Free during the beta |
Where Projects is the better choice
- You want Claude to split and coordinate the work. Projects has a coordinator that delegates to parallel threads and assembles the result. Memlo has no coordinator.
- All your agents are Claude, in the cloud. The shared memory comes with the project, with nothing to connect.
- You want each thread on its own branch. Every thread works on its own branch and copy of the repository.
- You're on Pro or Max and in the beta. It is available today to those subscribers.
Where Memlo is the better choice
- You use more than one vendor. Codex, ChatGPT and Gemini read and write the same memories as Claude Code.
- You run agents on your own computer. Local Claude Code and Codex sessions connect today.
- You want to read, correct and roll back what agents remember. Every memory is a titled note with a full history, and
restore_memorybrings back any version. - You want to see who changed what.
memory_historylists every version with the person and the connected app or agent that wrote it. - The memory has to outlast one project. Spaces aren't tied to a goal: the next project, and the next tool, search the same notes.
- You share with people. Teammates get the same memories through their own agents, as Viewers or Editors.
Can you use both?
Yes. Projects can use connectors, so a project's threads could also reach Memlo as an MCP server, though we haven't tested Memlo inside a Projects thread. A practical split: let the project's own shared memory carry what its threads learn about this one goal, and save the decisions you want every agent and teammate to find, later and in other tools, to Memlo.
Frequently asked questions
Does Claude Code Projects have shared memory?
Yes. Anthropic's announcement says "Every thread now adds to and draws from a shared memory." The threads are Claude Code cloud sessions in the same project.
Who can use Claude Code Projects?
On October 30, 2026, the redesigned Projects was "available in beta to select Claude Pro and Max subscribers who use cloud sessions in Claude Code and don't have any existing projects on the web or desktop." Team and Enterprise plans come later.
Can Codex or ChatGPT use a Claude Code project's memory?
The announcement doesn't mention other vendors' agents. Memlo is built for that case: Codex, ChatGPT and Claude Code connect to the same server.
How is this different from Claude Code agent teams?
Agent teams are an experimental Claude Code feature, off by default, where a lead session works with teammate sessions. Each teammate has its own context window. See Shared memory for Claude Code subagents and agent teams.
Sources
Checked on October 30, 2026:
- Projects redesigned: from folder to conversation, Anthropic, September 17, 2026
- Orchestrate teams of Claude Code sessions, Claude Code docs
- Memlo: the setup page, how it works and terms
Try Memlo
Memlo is invitation only for now. Join the waitlist and we'll email you when your invitation is ready.
Join the waitlist