PlugWiki
Memanto logo

Memanto MCP Server: Setup for Claude, Cursor and Other Apps

OfficialLocal
GetPyPImemanto-mcpmoorcheh-ai/memantoDocs

Gives Claude or Cursor long-term memory that persists across chats and tools.

This server comes from the Memanto project by Moorcheh. It runs on your computer, but the memories are stored in Moorcheh's cloud under your API key. A free tier of 100K operations a month is mentioned in the README.

Seven memory tools are on by default. Four agent-management tools only appear if you set MEMANTO_EXPOSE_ADMIN=true.

Good for

  • People who are tired of repeating preferences to their assistant
  • Teams that want several agents to share one memory

Works in

12 should work

Tap an app to see how to set it up and any known problems.

Install Memanto MCP

Claude Code

  1. 1

    Copy this prompt.

    Add the Memanto MCP server to Claude Code. Run this command:
    
    claude mcp add memanto -e MOORCHEH_API_KEY=<MOORCHEH_API_KEY> -- uvx memanto-mcp
    
    Ask me for MOORCHEH_API_KEY and only put the values in that config, nowhere else.
    
    Then check that it loaded and tell me if I need to sign in or restart.
  2. 2

    Paste it into Claude Code. It shows the change and asks before running it.

Requires uv (Python).

Built from the package data we hold. Not yet tested by us.

Tools

  • Saves one thing to long-term memory

    Stores a single fact, preference, decision, goal or instruction in your Memanto memory so any connected assistant can find it later.

    Try: “Remember that I prefer concise answers.”

    Write · remember

  • Saves up to 100 memories at once

    Stores many memories in one call, for example facts pulled out of a document.

    Try: “Save the key decisions from this meeting as separate memories.”

    Write · batch_remember

  • Searches your memory by meaning

    Finds stored memories that match a question by meaning, not exact words. The assistant is told to check here before asking you to repeat something.

    Try: “What do you remember about my travel preferences?”

    Network · recall

  • Shows the newest memories

    Returns the latest memories first, with no query needed.

    Try: “What did we decide recently?”

    Network · recall_recent

  • Shows what was known on a date

    Returns memories as they stood at a past date and time.

    Try: “What did you know about the project on 2025-11-01?”

    Network · recall_as_of

  • Shows what changed since a time

    Returns memories added or changed after a point in time.

    Try: “What is new in memory since yesterday?”

    Network · recall_changed_since

  • Answers a question from your memories

    Writes a grounded answer built from the memories stored for the agent.

    Try: “Based on memory, which editor do I use?”

    Network · answer

  • Creates a new memory space

    Opt-in. Creates a new memory space (agent) with its own namespace. Only available when admin tools are turned on.

    Try: “Create a memory space called research.”

    Write · create_agent

  • Lists memory spaces

    Opt-in. Lists every agent the API key can see.

    Try: “Which memory spaces exist?”

    Network · list_agents

  • Shows one memory space

    Opt-in. Looks up the details of one agent.

    Try: “Show details of the research agent.”

    Network · get_agent

  • Removes a memory space locally

    Opt-in. Deletes the agent's local metadata. The remote memories are kept until purged with the Memanto CLI or REST API.

    Try: “Delete the old test agent.”

    Delete · delete_agent

Read from the source code

Permissions

medium risk · 2 things to watch

Medium risk

The assistant can save new memories to the Moorcheh cloud and read them back, but the default tools cannot delete anything. Keep the admin tools off and do not let it store secrets.

What it can do

  • Read your data
  • Create and change things
  • Delete things
  • Read-only mode

Watch out for

  • Your memories are stored in the Moorcheh cloud

    Anything the assistant saves with remember is sent to Moorcheh under your key. Tell it not to store secrets or private data. github.com

  • Admin tools are off unless you turn them on

    create_agent, list_agents, get_agent and delete_agent appear only with MEMANTO_EXPOSE_ADMIN=true. delete_agent removes local metadata, not the remote memories. github.com

Security advisories

No published security advisories as of Nov 1, 2026.

Technical details
Turning on read-onlyNo read-only mode is documented. Every default install includes the remember tools.
Talks toMoorcheh cloud API
Sign-in and keysAPI key in the MOORCHEH_API_KEY environment variable of the MCP client config.
Scopes
  • Moorcheh API key

Questions

Is the Memanto MCP server official?

Yes. It ships in the Memanto project repository (moorcheh-ai/memanto) and is published by the same team.

Do I need an API key?

Yes. You need a Moorcheh API key from console.moorcheh.ai. The README says there is a free tier of 100K operations a month.

Where are my memories stored?

In the Moorcheh cloud, under your API key. The server itself runs on your computer.

Can it delete memories?

Not with the default tools. The opt-in delete_agent tool only deletes local metadata, and the remote memories stay until you purge them with the CLI or REST API.

Can I share memory between Claude and Cursor?

Yes. Agents that use the same key and the same agent ID read the same memory.

Which Python version does it need?

Python 3.10 or newer.

Information

updated Nov 2, 2026

Publisher
Official
Package
memanto-mcp
Registry
PyPI
Transport
stdio
Auth
API key
Version
0.1.5
License
MIT
Last commit
Oct 30, 2026
Open issues
5
Listed in
Official MCP registry

WebsiteDocsGitHubPyPIOfficial MCP RegistryChangelogIssues

Sources

Docs, issue trackers and discussions this page is built from. Issues, quotes and findings above link to their own source.

Show all 2
  1. Memanto MCP README · github.com
  2. memanto_mcp tools · github.com

Facts from the official MCP registry, GitHub, npm, PyPI and the Docker catalog. Updated Nov 2, 2026.