One memory. Every AI tool.
Persistent memory infrastructure for AI agents — write once, recall anywhere.
One API key; the same memory across Claude, ChatGPT, Cursor, VS Code, and any MCP client.
Mnemoverse stores preferences, decisions, lessons, and context — and serves them back to your agent in any tool. Memory that learns (Hebbian associations), consolidates (HDBSCAN clustering with Von Restorff protection, so distinctive memories survive compression), and improves with use (Rescorla-Wagner outcome feedback reranks what actually helped) — not just similarity search.
# Claude Code — one command
claude mcp add mnemoverse -e MNEMOVERSE_API_KEY=mk_live_YOUR_KEY -- npx -y @mnemoverse/mcp-memory-server@latest
# Python
pip install mnemoverseFree API key at console.mnemoverse.com — no credit card, 1,000 queries/day.
Point your agent at our machine-readable docs — API reference, per-tool setup, and pricing in one file (llms.txt standard):
https://raw.githubusercontent.com/mnemoverse/.github/main/llms.txt
| Repository | What it does |
|---|---|
| mcp-memory-server | MCP server for Claude Code, Cursor, VS Code, Windsurf — and any MCP client |
| mnemoverse-vscode | VS Code extension — persistent memory for AI coding assistants |
| mnemoverse (PyPI) | Python SDK — sync + async clients, retry & circuit breaker |
| mnemoverse-slod-paper | Code companion to the SLoD paper (arXiv:2603.08965) |
- Semantic Level of Detail (SLoD) — published on arXiv:2603.08965
- Benchmarked on public memory benchmarks — LoCoMo, LongMemEval, HotpotQA
- 7 provisional patents filed (INPI, 2026)
- NVIDIA Inception Program member
- Documentation — guides, API reference, integration setup
- Console — sign up, API keys, usage dashboard
- Research library — the science behind the memory layer
- ChatGPT setup — Custom GPT with Actions
