Mnemoverse Memory

Persistent memory for AI agents, shared across tools. One account gives Claude Code the same long-term memory it has in Cursor, VS Code, and any other MCP client: write a memory in one tool, recall it in another.

This plugin connects the hosted Mnemoverse memory server over MCP and adds the discipline that makes memory actually get used: recall before acting, save durable decisions and corrections afterwards.

Install

From the community marketplace (once the plugin is accepted there):

claude plugin marketplace add anthropics/claude-plugins-community
claude plugin install mnemoverse@claude-community

Or try it directly from this repository:

claude --plugin-dir ./

First run: sign in once, no API key

The server at https://mcp.mnemoverse.com/mcp uses OAuth 2.1 with PKCE. On first use Claude Code opens a browser window; sign in to your Mnemoverse account (free tier at console.mnemoverse.com, no credit card) and grant the requested scopes, which include memory:read and memory:write. There is no API key to paste, and access can be revoked at any time from the console. The bundled setup skill walks Claude through this if anything goes wrong.

What you get

Three commands:

Command	What it does
/mnemoverse:remember	save a durable fact, decision, or correction, one fact per memory
/mnemoverse:recall	search memory before acting, with provenance, surfacing contradictions instead of hiding them
/mnemoverse:memory-status	counts, domains, and the newest memories

Two skills:

• agent-memory-discipline (CC0, backend-neutral): standing rules for when to recall and what is worth saving. It names no vendor and works on any memory backend; it ships here so the connected memory is actually used.
• setup: guides Claude through connecting and troubleshooting the MCP server.

Ten tools from the remote server: memory_read, memory_write, memory_list_recent, memory_stats, memory_feedback, four shared-room tools (memory_create_room, memory_invite_to_room, memory_join_room, memory_list_rooms), and vault_list (aliases only, values are never returned).

The two delete tools available in the local package are deliberately not exposed over the remote connector, so a one-click sign-in can never wipe memory.

Try these

1. Remember that this project uses pnpm, not npm. We decided this after a lockfile conflict on 2026-08-01.
2. Before you touch the billing module, recall what we have decided about it in past sessions.
3. /mnemoverse:memory-status after a week of use, to see what your sessions have been saving.

What gets stored, and what does not

The plugin stores only what you or Claude explicitly save through the memory tools: single facts, decisions, corrections. It does not record conversations, does not read your chat history, and the bundled skill explicitly refuses to persist secrets, API keys, passwords, payment data, or personal identifiers.

Memory content lives in your Mnemoverse account. Privacy policy: mnemoverse.com/privacy.

Troubleshooting

• No browser window on first use: run /mcp to check server status, or re-add manually with claude mcp add --transport http mnemoverse https://mcp.mnemoverse.com/mcp.
• 401 after it used to work: the token expired; the next tool call normally refreshes it. If not, remove and re-add the server to re-run sign-in.
• malformed_token in logs: something is sending an API key as a bearer token. The remote endpoint takes OAuth tokens only; API keys are for the REST API and the local package.

More: docs.

Support

Issues in this repository, or helloworld@uinside.org.

License

Plugin: MIT. The bundled agent-memory-discipline skill is CC0-1.0 and lives at mnemoverse/agent-memory-discipline.