M3 Memory runs on your own machine — the client starts it, so there is no endpoint to ping. 1 480 installs a week from pypi. Last commit 26 Oct 2026.
Local-first, cloud-friendly memory — 100+ tools, 99.2% LongMemEval-S retrieval@8, hybrid search.
We read the source, 1 h ago · rules 995398eba69f
A value the model can set ends up inside a file or shell call. That is not a flaw by itself — for a terminal server it is the job — but it is where things go wrong when it is not.
return subprocess.run(cmd, timeout=timeout).returncode # nosec B603 - argv list, no shell
What this server is able to do. For an MCP server this is often the job itself — a terminal server runs commands because that is what it is for. Listed so you know what you are plugging in, not as an accusation.
этот файл ставится пользователю, но в репозитории его нет
_LAUNCHD_PLIST = "~/Library/LaunchAgents/com.m3memory.cognitiveloop.plist"
out = subprocess.run(["ifconfig", iface], capture_output=True, text=True, timeout=2).stdout
candidates.append(os.path.join(m3lib, nm))
pg_cur.execute(select + " WHERE wh_arrived_at > %s" + kind_sql, (watermark, *kind_params))
exec(code, {"__file__": str(bridge), "__name__": "__main__"}) # nosec B102
Is this your server and something here is wrong? Tell us — corrections are free and do not require a plan.
That is not a flaw by itself — but it is where things go wrong when it is not the job. We re-read this code on every release. Watch it and you hear from us the day another one appears.
This server runs on your own machine — install it with the package manager and the client starts it for you. Package name taken from the official registry entry.
claude mcp add m3-memory -- uvx m3-memory
{
"mcpServers": {
"m3-memory": {
"command": "uvx",
"args": [
"m3-memory"
]
}
}
}
[mcp_servers.m3-memory]
command = "uvx"
args = ["m3-memory"]
{
"mcpServers": {
"m3-memory": {
"command": "uvx",
"args": [
"m3-memory"
]
}
}
}
{
"servers": {
"m3-memory": {
"type": "stdio",
"command": "uvx",
"args": [
"m3-memory"
]
}
}
}
Local-first MCP code intelligence: 37 tools — hybrid search, blast-radius, diff review, memory.
Persistent memory for AI tools. Local-first knowledge graph with hybrid search.
Local encrypted memory for AI: hybrid search, 8 MCP tools, one file you own. Free, no cloud.
Markdown-first long-term memory for AI agents. Hybrid BM25 + vector search, runs locally.
Long-term memory for AI agents: local-first PKM with provenance and hybrid search.
Local-first memory for Claude Code and any MCP client: hybrid search + knowledge graph, $0/token.
Code indexing MCP: 15 tools, 10 languages, hybrid search, call graphs, O(1) retrieval.
Code indexing MCP: 15 tools, 10 languages, hybrid search, call graphs, O(1) retrieval.
Answers built from our own checks of this server.