Memanto MCP Server

Name: Memanto MCP Server
Author: Moorcheh.ai

Typed, long-term memory for MCP clients, stored in Moorcheh's semantic memory service

Official · Moorcheh.ai Python Local · stdio Knowledge and memory

GitHub PyPI Website

38 popularity score ★ 2.4k ↓ 313/mo

Connect your client

Local (PyPI)

Add in one click

Add to Cursor Add to VS Code Add to LM Studio

Opens the app, which asks before it adds the server. Then replace each <NAME> in its settings with your value.

Or add it by command or config

It needs MOORCHEH_API_KEY. Replace each <NAME> with your value.

FirstTerminal

pip install memanto-mcp

Claude CodeTerminal

claude mcp add memanto -e MOORCHEH_API_KEY=<MOORCHEH_API_KEY> -- memanto-mcp

Claude Desktopclaude_desktop_config.json

{
  "mcpServers": {
    "memanto": {
      "command": "memanto-mcp",
      "args": [],
      "env": {
        "MOORCHEH_API_KEY": "<MOORCHEH_API_KEY>"
      }
    }
  }
}

Cursor~/.cursor/mcp.json

{
  "mcpServers": {
    "memanto": {
      "command": "memanto-mcp",
      "args": [],
      "env": {
        "MOORCHEH_API_KEY": "<MOORCHEH_API_KEY>"
      }
    }
  }
}

VS Code.vscode/mcp.json

{
  "servers": {
    "memanto": {
      "type": "stdio",
      "command": "memanto-mcp",
      "args": [],
      "env": {
        "MOORCHEH_API_KEY": "<MOORCHEH_API_KEY>"
      }
    }
  }
}

Codex~/.codex/config.toml

[mcp_servers.memanto]
command = "memanto-mcp"
args = []

[mcp_servers.memanto.env]
MOORCHEH_API_KEY = "<MOORCHEH_API_KEY>"

Gemini CLITerminal

gemini mcp add --scope user -e MOORCHEH_API_KEY=<MOORCHEH_API_KEY> memanto memanto-mcp

Settings 10

`MOORCHEH_API_KEY`	Required: Moorcheh API key; the server will not start without it.
`MEMANTO_DEFAULT_AGENT_ID`	Optional: Agent used when a tool call gives no agent_id; recommended, one per project.
`MEMANTO_EXPOSE_ADMIN`	Optional: Set to true to add the four agent-management tools (default false).
`MEMANTO_AGENT_AUTO_CREATE`	Optional: Makes the default agent on first use if it is missing (default true).
`MEMANTO_AGENT_PATTERN`	Optional: Pattern for an auto-created agent: support, project or tool (default tool).
`MEMANTO_SESSION_DURATION_HOURS`	Optional: Session lifetime in hours, 1 to 720 (service default 6).
`MEMANTO_MCP_TRANSPORT`	Optional: stdio (default), sse or streamable-http.
`MEMANTO_MCP_HOST`	Optional: Bind address for sse/http (default 127.0.0.1).
`MEMANTO_MCP_PORT`	Optional: Bind port for sse/http (default 8765).
`MEMANTO_MCP_LOG_LEVEL`	Optional: Log level, written to stderr (default INFO).

What it does

memanto-mcp is the MCP part of the Memanto project from Moorcheh. It gives a client such as Claude Desktop, Cursor or Windsurf a memory that lasts across chats. The assistant saves short notes, each tagged with one of 13 types (fact, preference, decision, goal, instruction, error and others). Later it searches them by meaning, lists the newest ones, asks what was known on a given date or what changed since then, or gets an answer written from the saved notes.

Each memory belongs to an agent, a named store. Clients that use the same agent id share one memory, and the server tags each write with the name of the client that made it, so you can tell later who wrote what. The first call makes the default agent if it does not exist yet (MEMANTO_AGENT_AUTO_CREATE, on by default). Seven memory tools are always on. Four agent-management tools appear only when MEMANTO_EXPOSE_ADMIN=true.

The main Memanto README describes a fully local mode (Docker plus Ollama, no account), but this MCP package does not use it. Its settings file (integrations/mcp/memanto_mcp/config.py) refuses to start without MOORCHEH_API_KEY, and its own README says calls go to the Moorcheh service over HTTPS. Searches return 10 memories by default, up to 100 with limit, and batch_remember takes at most 100 items per call (tools.py).

What it needs

Runtime	Python 3.10+

Python 3.10 or newer, with memanto-mcp installed by pip (it brings the memanto package and the MCP SDK). A Moorcheh API key from console.moorcheh.ai; the free tier allows 100K operations a month. An MCP client that can start a local stdio command, or one that connects over SSE or streamable HTTP if you start the server that way.

Limits

Needs a Moorcheh cloud key: the local Docker/Ollama mode in the main Memanto README is not one of this server's settings. Searches return 10 memories by default, up to 100. batch_remember takes up to 100 items. Session lifetime can be set from 1 to 720 hours (default 6). delete_agent removes only the agent's local metadata; the memories stay in Moorcheh until you purge them with the Memanto CLI or REST API. The policy, conflict-review, import and export features in the main README are CLI commands, not MCP tools. Over SSE or HTTP the server does not check who connects; its docs say to put a reverse proxy with auth in front.

Who it suits

Suits people who want one memory shared across several MCP clients or projects, with typed notes and date-based recall, and who are fine with that memory living in Moorcheh's hosted service. Pick something else if memory must stay on your own machine or offline: this server always calls Moorcheh, even though the Memanto CLI has a local mode. Leave the admin tools off unless the assistant should be able to create or remove agents.

Tools 11

remember: Saves one memory with a type, an optional title, tags, a confidence score and a source
batch_remember: Saves up to 100 memories in one call
recall: Searches memories by meaning, with filters for type and score; 10 results by default, up to 100
recall_recent: Lists the newest memories first, with no search query
recall_as_of: Shows what the memory held on a given date
recall_changed_since: Lists memories added or changed since a given date
answer: Writes an answer to a question using only the stored memories, and lists the ones it used
create_agent: Makes a new agent (memory store); only with MEMANTO_EXPOSE_ADMIN=true
list_agents: Lists the agents the API key can see; only with MEMANTO_EXPOSE_ADMIN=true
get_agent: Shows one agent's details; only with MEMANTO_EXPOSE_ADMIN=true
delete_agent: Removes an agent's local metadata but keeps its stored memories; only with MEMANTO_EXPOSE_ADMIN=true

Popularity

38 popularity score, on our list (the bar is 36)

★ 2.4k ↓ 313/mo

GitHub stars, whole repository	2,357	15.2 points
PyPI downloads, last month	313	15.2 points
Official server	Yes	8 points

It lives in the folder integrations/mcp of moorcheh-ai/memanto, which holds more than this server. Its stars are the whole repository's, so in the score they never count for more than its own downloads.

Measured 2026-11-07. How the score works.

Where it is published

Code	moorcheh-ai/memanto/integrations/mcp ★ 2.4k, the whole repository
PyPI	memanto-mcp 0.1.5, 2026-10-29

What it can reach

Runs	On your machine
Reaches	The internet, Your account at a service
Changes things	Yes: it can write, send, create or delete
Sign-in	An API key or token
Licence	MIT (GitHub)
Last code change	2026-10-30

Runs on your machine. Every tool sends memory text to Moorcheh's hosted service over HTTPS, signed in with your MOORCHEH_API_KEY, so what you store lives on Moorcheh's servers. Tools can write: remember and batch_remember add memories. With MEMANTO_EXPOSE_ADMIN=true the assistant can also create agents and remove agent metadata, and list every agent the key can see. The server does not read local files or run commands. By default it uses stdio. In sse or streamable-http mode it binds to 127.0.0.1 unless you change the host, and it does not check who connects, so anyone who can reach the port can use your key's memory. Settings can also come from a .env file in the working directory.

Over time

GitHub stars, whole repository 2.4k +35 (+2 %)

2.4k2.3k

15 Oct 20267 Nov 2026

PyPI downloads a month 313 +137 (+78 %)

431124

22 Jun 20267 Nov 2026

Before the site measured it, read from pypistats' daily download counts.

Popularity score 38 +0

3838

15 Oct 20267 Nov 2026

What is rising across the directory

Sources

Page checked 2026-10-15.

Badge for your README

Shows this server's rank in its category, and stars for a server that runs on your machine. It updates every hour.

MarkdownREADME.md

[![Memanto MCP Server on EveryMCPServer](https://everymcpserver.com/badge/memanto.svg)](https://everymcpserver.com/servers/memanto)

Connect your client

Add in one click

Or add it by command or config

Settings 10

What it does

What it needs

Limits

Who it suits

Tools 11

Similar servers

Commonly MCP Server

State Memory MCP

Remnic

Claude-Mem

Anytype MCP Server

Compartment