MEMESH

A local collaboration layer for AI coding agents

MeMesh lets AI coding agents share project memories, exchange questions and replies, and hand over work. Let Codex implement and Claude Code review, then send feedback to the original conversation to keep the discussion moving. Free and open source, with memories and collaboration records stored on your computer.

Work across conversations
Discuss, divide tasks and hand over work between tools and sessions
Share memories
Give Claude Code, Codex and Hermes Agent the same project context
Keep it local
Your memories and collaboration records stay under your control

WHEN YOU USE MORE THAN ONE AI CONVERSATION

Several AI assistants, but you are still passing every message

  • Copying messages between Codex and Claude Code so they can work together
  • Explaining the task and progress again before another conversation can review it
  • Repeating project rules and preferences in every new chat
  • Going in circles on a problem you already solved

WHAT MEMESH HELPS YOU DO

Shared memory and two-way conversations that keep development moving

01

Collaborate across tools and conversations

Ask Codex to implement and Claude Code to review. Send questions back to the original conversation to keep discussing and refining the work. Different conversations in the same tool can collaborate too.

  • Choose the conversation to exchange questions, replies and results with
  • Keep messages and handling records for the next step

02

Switch tools without repeating yourself

Connect different tools to MeMesh to use the same project memories. A decision saved in one tool can be found from another.

  • Share project context across Claude Code, Codex and Hermes Agent
  • Connect compatible tools such as Cursor and Gemini CLI too

03

Pick up in a new conversation

Keep project decisions, progress and next steps. Find the relevant memories when you return, so you spend less time starting from scratch.

  • Save your preferences and project rules
  • Leave a handoff that helps the next conversation get up to speed

04

Put past fixes to work again

Keep the problem, its cause and the fix for later reference. The Claude Code plugin also brings in relevant memories before file edits.

  • Keep useful fixes and lessons from past work
  • Make a solution you already found easier to use next time

GET STARTED

Give your project a memory, then let conversations work together

  1. 01

    Install MeMesh

    Follow the official guide to add the plugin to Claude Code or Codex.

  2. 02

    Save the project context

    For example: “Remember that this project uses Traditional Chinese for its interface.”

  3. 03

    Ask two conversations to collaborate

    For example: “Have Codex send this change to Claude Code for review and return the feedback.”

  4. 04

    Keep the work moving

    Continue with the reply in your original conversation, or find earlier decisions and handoffs in a new one.

See how to install →

No MeMesh account needed. Start on your own computer.

FREE, OPEN SOURCE AND LOCAL

Your memories stay under your control

Free and open source

MIT licensed, with public source code. Install it, use it for free, and inspect how it works.

No extra AI API needed

Store and search memories locally without signing up for a separate MeMesh account or API key.

Update, archive and export

Update a memory when a decision changes, archive what you no longer need, or export a copy.

BEFORE YOU TRY IT

Common questions

Who is MeMesh for?

People using AI coding agents who want them to share project context and collaborate. It is especially useful when you divide implementation and review between agents or continue work in a new conversation.

What can it remember?

Your preferences, project rules, decisions, useful fixes and where work stopped. You can ask the AI to save a memory or find an earlier one in plain language.

Which AI tools can I use?

Claude Code and Codex can share memories and collaborate across conversations. Hermes Agent also has a native memory integration. Compatible tools such as Cursor and Gemini CLI can connect through the integration guide.

How do different conversations collaborate?

Follow the integration guide to connect MeMesh to your tools, then ask your AI to find a conversation in the same project and send it a question or task. Claude Code and Codex can exchange replies and results; separate conversations in the same tool can collaborate too.

Do I need a cloud account or another AI API?

No. MeMesh stores and searches memories on your computer, without a separate MeMesh account or model API. Your existing AI coding tool keeps its own setup.

Can I correct a memory or stop using it?

Yes. Update memories as decisions change, archive what you no longer need, or export a copy. The data stays on your computer.

Keep the memory. Connect the conversations.

Let the AI coding agents you use share project context, discuss the work and divide tasks, with less repeated explaining and copying messages between conversations.

MeMesh is built and open-sourced by Partners Circle: the local collaboration layer our own team uses for AI coding agents.