M3 Memory MCP Server

A memory layer that outlives your agents. You switch from Claude Code to Cursor, upgrade your model, start fresh next week - and everything your tools learned about your project is gone. You re-explain the same decisions, the same preferences, the same hard-won context, over and over. It runs locally over stdio via the published package.

People who work with agentic memory, gdpr and gemini cli and want it reachable from Claude, Cursor, VS Code, or another MCP client. The project is written in Python.

VERIFIED ACTIVE

LAST COMMIT 2026-09-18 · ★ 25 · #18 OF 304 MAINTAINED MEMORY · VERIFIED 2026-09-18

Apache-2.0 · Python servers · how we verify → /methodology

01 · Install M3 Memory

before you install - you'll need

The README does not document required environment variables for a basic install.

Claude Code

claude mcp add skynetcmd-m3-memory -- uvx m3-memory

Claude Desktop / Cursor / VS Code - add to config

{
  "mcpServers": {
    "skynetcmd-m3-memory": {
      "command": "uvx",
      "args": [
        "m3-memory"
      ]
    }
  }
}

Same JSON for Cursor. For VS Code, rename the top-level key from `mcpServers` to `servers`.

Using another client? Same JSON, different key

Claude Desktop · mcpServers

Cursor · mcpServers

VS Code · servers

Windsurf · mcpServers

Zed · context_servers

Cline · mcpServers

Roo Code · mcpServers

Continue · mcpServers

LibreChat · mcpServers

Gemini CLI · mcpServers

Codex CLI · mcp_servers

Full setup guides: every client.

02 · Evidence

Security posture

What to check before giving this server access to your agent - from the registry, GitHub, and our own probes. We don't score safety; we show what's verifiable.

runs as local process (stdio) - runs on your machine with your user's permissions

license Apache-2.0 - declared in the repository

pypi package m3-memory - check the name against the project README before installing (PyPI has no namespace ownership)

registry namespace io.github.skynetcmd is GitHub-verified and matches the repo owner

03 · What M3 Memory can do

Prose above is summarized from the project's README and registry record - no invented capabilities.

Latest releases

v2026.9.17.0 · 2026-09-18

Changed · Startup registers 10 tool schemas instead of 20 - about 2% of a 200K · context window, down from 3%. Every other tool is reached as before, through · tools_load_domain or by name with m3_call; M3_TOOLS_LAZY=0…

v2026.9.16.0 · 2026-09-17

Added · The files subsystem reads HTML, Word, PowerPoint, Excel, RTF, EPUB, · OpenDocument and Apple iWork documents. .html/.htm/.xhtml split on · their own headings with scripts, styles and markup removed; .docx…

v2026.9.14.4 · 2026-09-15

Fixed · notifications_poll no longer returns claimed rows. An in-flight message was · offered to every poller, so two agents could do the same work. The unread path · now selects PENDING through message_state_sql; the…

04 · Who maintains M3 Memory

M3 Memory is maintained by skynetcmd. It's the only MCP server we track from this author; the repo dates to Feb 2026.

05 · Facts

category
memory - ranked #18 of 304 actively-maintained memory servers as of 2026-09-18.
release cadence
10+ releases in the last 90 days (latest 2026-09-18)
registry
io.github.skynetcmd/m3-memory (active, first published 2026-04-10 · 92 versions)
packages
pypi:m3-memory

06 · M3 Memory FAQ

What is M3 Memory?

A memory layer that outlives your agents. You switch from Claude Code to Cursor, upgrade your model, start fresh next week - and everything your tools learned about your project is gone. You re-explain the same decisions, the same preferences, the same hard-won context, over and over. It runs locally over stdio via the published package.

Is M3 Memory still maintained?

Yes - as of 2026-09-18, its last commit was 2026-09-18 and it shipped 10+ releases in the last 90 days. We re-verify nightly.

How do I install M3 Memory?

Run `uvx m3-memory`. The README does not document required environment variables for a basic install. You can also paste the ready-made client config above.

Does M3 Memory run locally?

Yes - it's a stdio server: it runs on your machine (via uvx) with your user's permissions. Your data stays local unless the server itself calls external APIs.

07 · Alternatives to M3 Memory

Pairs well with

Servers that cover what M3 Memory doesn't - only shown when the pairing reason fits the companion.

  1. Server Filesystem → filesystem companion filesystem · ★ 39,038
  2. Pdf MCP → filesystem companion filesystem · ★ 135