MnemoverseDocs

Appearance

MnemoverseOne Memory. Every AI Tool.

Your AI remembers — preferences, decisions, lessons — across Claude, ChatGPT, Cursor, and other connected tools. One account, shared memory. Choose the setup for your client.

Get Started Free
For AI Agents

Claude Code & Claude apps

One command for the Claude Code CLI, or a one-click connector for the Claude Desktop, web, and mobile apps. Memory persists across every session.

Setup guide

ChatGPT

Persistent memory in ChatGPT. Connect over MCP with a browser sign-in, or build a Custom GPT on a workspace plan.

Setup guide

Cursor, VS Code & Windsurf

JSON config per editor. Same memory as Claude and ChatGPT — write in one, recall in any other.

Setup guide

Python SDK & REST API

pip install mnemoverse — 3 lines to add memory to any agent, bot, or backend.

SDK docs

Rooms — Shared Memory Across Accounts (Beta)

One memory pool shared by several accounts — your team's agents, or your Claude and a colleague's ChatGPT. Membership-checked on every call; each member's usage billed to their own account.

Rooms guide

Smart Memory, Not Storage

Hebbian associations link related memories. Feedback loop learns what's useful. Consolidation is designed to merge duplicates and is off on the hosted service. Memory gets better over time.

Built on Research

SLoD published on arXiv. Benchmarked on LoCoMo (1,986 questions) and LongMemEval — see the benchmarks page for current numbers.

See benchmarks

How It Works ​

text
                    ┌── Claude Code (MCP)
                    ├── Claude Desktop (MCP)
                    ├── Cursor / VS Code (MCP)
   Mnemoverse API ──├── ChatGPT (MCP / Actions)
   (one memory)     ├── Python SDK
                    └── REST API / Your app

Write a memory in Claude Code. Read it in ChatGPT. Every tool connected as the same Mnemoverse account shares one memory, whether it signs in through the browser or uses an API key.

Need to share beyond your own account? A room (Beta) is a memory pool shared across accounts — members address it as xroom:<room_id> in the same write/read calls.

Quick Start ​

Option 1: Give your agent the docs (llms.txt standard)

https://mnemoverse.com/docs/llms.txt

Option 2: One command (Claude Code)

bash
claude mcp add mnemoverse -s user \
  -e MNEMOVERSE_API_KEY=mk_live_YOUR_KEY \
  -- npx -y @mnemoverse/mcp-memory-server@latest

Option 3: Hosted connector with browser sign-in (Claude Desktop / claude.ai, ChatGPT developer mode, editors)

Add a custom connector with this URL — OAuth sign-in, no key to paste (how):

text
https://mcp.mnemoverse.com/mcp

Option 4: REST API (direct HTTP for curl/SDK — not an MCP connector URL)

bash
curl -X POST https://core.mnemoverse.com/api/v1/memory/write \
  -H "X-Api-Key: mk_live_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"content": "User prefers Railway for deploys", "concepts": ["deploy", "railway"]}'

Using an API-key setup? Get your free API key at console.mnemoverse.com. For browser sign-in without pasting a key, create a free account at the same link, then use Option 3 above.

Learn More ​

From the Library ​

Field notes on AI agent memory — start here, then browse the full Library: