Skip to content
verify mcp Beta VerifyMCP is currently in beta. If you notice any issues, get in touch and we’ll put it right.

M3 Memory

PYPI · M3-MEMORY · SCANNED SEP 22

Local-first memory — 100+ tools, 99.2% LongMemEval-S retrieval@10, hybrid search, GDPR, no cloud.

Available components

−8 this week 53 Trust /100
Trust breakdown (7 categories)

How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. How we score → Why this is hard to score →

Supply Chain Security49
  • Malware scan not yet available for this package.Unverified
  • No known CVEs affecting this package version or its production dependencies.Pass
  • Runs setuptools.build_meta at install time, a recognised native-build step with no shell scripting around it. View diagnostics → Pass
  • 10 of 85 dependencies flagged as unhealthy. View diagnostics → Partial
Provenance & Transparency35
Schema Quality & AI Usability57
  • AI-judged instruction clarity (good).Pass
  • Context-footprint check failed: tool/resource definitions use about 1810 tokens (~181/item across 10 items; 10 tools + 0 resources), over budget; trim descriptions and params. See how to fix → Fail
  • Usage-examples check failed: none of the tools include examples. See how to fix → Fail
Stability & Change Management40
  • Stability check failed: the tool surface changed between 2026.8.19.16 and 2026.9.21.0: 10 tool removals, 0 breaking changes, 0 additions. See how to fix → Fail
Tool Coverage67
  • 100% of tools have a non-trivial description (not blank, and not just the tool's name).Pass
  • 0% of tool parameters carry a description.Fail
Tool Safety100
  • No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.Pass
  • We read all 10 captured tool definition(s), and no name or description among them implies an irreversible operation.Pass
  • An AI judge read all 11 captured unit(s) of tool text and found none that tries to manipulate the model reading it.Pass
Capabilities100
  • Implements a supported MCP spec version (2025-11-25); the latest is 2026-07-28.Pass
Install

How do I install the M3 Memory MCP server?

M3 Memory runs locally as a PyPI package, launched with uvx m3-memory. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.

pypi · m3-memory

# add to Claude Code
claude mcp add skynetcmd-m3-memory -- uvx m3-memory
// .cursor/mcp.json
{
  "mcpServers": {
    "skynetcmd-m3-memory": {
      "command": "uvx",
      "args": [
        "m3-memory"
      ]
    }
  }
}
// .vscode/mcp.json
{
  "servers": {
    "skynetcmd-m3-memory": {
      "command": "uvx",
      "args": [
        "m3-memory"
      ]
    }
  }
}
# add to Codex CLI
codex mcp add skynetcmd-m3-memory -- uvx m3-memory
// opencode.json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "skynetcmd-m3-memory": {
      "type": "local",
      "command": [
        "uvx",
        "m3-memory"
      ],
      "enabled": true
    }
  }
}
# add to OpenClaw
openclaw mcp add skynetcmd-m3-memory --command uvx --arg m3-memory
# ~/.hermes/config.yaml
mcp_servers:
  skynetcmd-m3-memory:
    command: "uvx"
    args: ["m3-memory"]
// ~/.netclaw/config/netclaw.json
{
  "McpServers": {
    "skynetcmd-m3-memory": {
      "Transport": "stdio",
      "Command": "uvx",
      "Arguments": [
        "m3-memory"
      ]
    }
  }
}
# add to Vellum
assistant mcp add skynetcmd-m3-memory -t stdio -c uvx -a m3-memory
// mcp.json
{
  "mcpServers": {
    "skynetcmd-m3-memory": {
      "command": "uvx",
      "args": [
        "m3-memory"
      ]
    }
  }
}
Changelog

Every change we have recorded for this component, newest first. Security-relevant changes are always shown. ▲ marks a change for the better, ▼ a change for the worse; unmarked changes are neutral.

  • 22 Sept 26 +16
    • Stability: unverified → fail security
    • Malware scan: pass → unverified security
    • Injection markers: unverified → pass security
    • Tool coverage: unverified → 100 functional
    • MCP protocol: unverified → pass functional
    • Package version: 2026.9.20.3 → 2026.9.21.0 functional
  • 21 Sept 26 0
    • Package version: 2026.9.17.0 → 2026.9.20.3 functional
  • 20 Sept 26 +4
    • Stability: unverified → fail security
    • Injection markers: unverified → pass security
    • Known CVEs: partial → pass security
    • Tool coverage: unverified → 100 functional
    • MCP protocol: unverified → pass functional
    • Dependency health: partial → 0.95 functional
    • Package version: 2026.9.19.0 → 2026.9.20.1 functional
    • Package version: 2026.9.19.0 → 2026.9.20.0 functional
    • Package version: 2026.9.17.0 → 2026.9.19.0 functional
  • 19 Sept 26 −18
    • Stability: fail → unverified security
    • Tool safety: pass → unverified security
    • Known CVEs: pass → partial security
    • Malware scan: unverified → pass security
    • Capabilities: pass → unverified functional
    • Tool coverage: 100 → unverified functional
    • Dependency health: 0.95 → partial functional
    • Package version: 2026.9.17.0 → 2026.9.19.0 functional
  • 18 Sept 26 +14
    • Stability: unverified → fail security
    • Malware scan: pass → unverified security
    • Injection markers: unverified → pass security
    • Schema quality: 129 → 181 functional
    • Tool coverage: unverified → 100 functional
    • MCP protocol: unverified → pass functional
    • Package version: 2026.9.16.0 → 2026.9.17.0 functional
  • 17 Sept 26 −40
    • Stability: 0.97 → unverified security
    • Tool safety: pass → unverified security
    • Malware scan: pass → unverified security
    • Stability: 0.97 → pass security
    • Capabilities: pass → unverified functional
    • Tool coverage: 100 → unverified functional
    • Package version: 2026.9.14.4 → 2026.9.16.0 functional
  • 16 Sept 26 +16
    • Malware scan: unverified → pass security
  • 15 Sept 26 +24
    • Malware scan: pass → unverified security
    • Injection markers: unverified → pass security
    • Tool coverage: unverified → 100 functional
    • MCP protocol: unverified → pass functional
    • Stability: unverified → 0.93 functional
    • Package version: 2026.9.14.3 → 2026.9.14.4 functional
Diagnostics

Diagnostic detail from the automated scan of this channel: what the scanner observed at each step, so you can see exactly where a check passed or failed. It is informational only and never changes the trust score.

Captured 22 Sept 2026 · Analysed pypi/m3-memory@2026.9.21.0

Provenance No attestation

The registry publishes no build provenance for this version, so there is nothing to verify.

Result No attestation
Ecosystem pypi

Background: How many MCP packages publish verified provenance →

Install scripts 1 script
Hook Tier Command
build_backend allowlisted setuptools.build_meta

Background: Why install scripts are a supply-chain risk →

Dependencies 85 packages
Packages resolved 85
Stale 10
Tree resolution Complete

Background: SBOMs and build attestations, explained →

MCP tools · 10 exposed · ~1,467 tokens

The tools this component advertises to a client, with an estimated token cost for each. Expand a tool to see its parameters and schema. The per-tool counts are indicative and are not scored directly; the schema's total context footprint is one signal in Schema Quality & AI Usability. A tool's description is untrusted text the model reads on every call, which is what makes this list a security surface and not just an inventory: how tool poisoning works →

Tool Tokens
chatlog_search ~118

Search chat_log rows. FTS5 keyword when query is non-empty; filter-only when empty.

NameTypeReqDescription
agent_idstring
conversation_idstring
databasestring
host_agentstring
kinteger
model_idstring
providerstring
querystringyes
search_modestring
sincestring
timeoutnumber
untilstring

No output schema declared.

No examples provided.

chatlog_status ~59

One-call health summary of the chat log subsystem: mode, DB paths, row counts, queue depth, spill files, embed backlog, hook timestamps, redaction state, warnings.

NameTypeReqDescription
databasestring
timeoutnumber

No output schema declared.

No examples provided.

files_search ~109

Hybrid FTS5 + vector search over file-ingestion leaves. Default: current versions only. Set include_history=True for time-travel queries. Use `corpora` for fan-out across multiple corpora.

NameTypeReqDescription
corporaarray
corpusstring
databasestring
filetypestring
include_historyboolean
limitinteger
querystringyes
timeoutnumber

No output schema declared.

No examples provided.

m3_call ~176

Invoke ANY m3 catalog tool by name without loading its domain — the low-token path to the full tool surface. Single call: pass `tool` (e.g. 'files_stats') and `args` (an object). Batch: pass `batch`, a list of {tool, args} (each isolated — one failure won't abort the rest; capped at 100). Set `dry_run` to validate args + check the destructive gate WITHOUT executing. Returns JSON. Call `m3_index` first if you don't know a tool's args. Destructive tools require MCP_PROXY_ALLOW_DESTRUCTIVE=1.

NameTypeReqDescription
argsobject
batcharray
databasestring
dry_runboolean
timeoutnumber
toolstring

No output schema declared.

No examples provided.

memory_get ~50

Retrieves a full MemoryItem; accepts full UUID or 8-char prefix; ambiguous prefixes return an error.

NameTypeReqDescription
databasestring
idstringyes
timeoutnumber

No output schema declared.

No examples provided.

memory_search ~178

Search across memory items using semantic similarity or keyword matching. Filter by user_id and scope for isolation.

NameTypeReqDescription
adaptive_kboolean
agent_filterstring
as_ofstring
as_recordsboolean
conversation_idstring
databasestring
explainboolean
include_bench_databoolean
include_scratchpadboolean
kinteger
querystringyes
recency_biasnumber
requesting_agentstring
scopestring
search_modestring
timeoutnumber
type_filterstring
user_idstring
variantstring

No output schema declared.

No examples provided.

memory_supersede ~386

Explicitly supersede an existing memory with a new one. Use this to record an intentional update — 'this fact replaces that specific memory' — when you know the old memory's id. Unlike memory_write's automatic contradiction detection (a cosine + title heuristic that may link the wrong prior memory or none at all), this targets the given old_id deterministically. Non-destructive: the old memory is retained, its validity interval is closed (is_deleted=1, valid_to set), and a 'supersedes' edge is recorded new -> old. The old memory stays retrievable by id and via memory_history, and as_of-filtered search still sees it valid before the supersession point — it is only dropped from default search. Fields you omit (type, title, importance, scope) are inherited from the old memory, so pass only what changed. To hard-delete instead, that is a separate gated tool (memory_delete). old_id MUST be the full UUID — a prefix is rejected (full UUID required for mutation safety; memory_get accepts a prefix, this does not). Note: each supersede creates a NEW successor memory; call it once with the full id, do not chain supersedes.

NameTypeReqDescription
agent_idstring
change_agentstring
contentstringyes
databasestring
embedboolean
embed_textstring
importancenumber
metadatastring
model_idstring
old_idstringyes
scopestring
sourcestring
timeoutnumber
titlestring
typestring
user_idstring
valid_fromstring
variantstring

No output schema declared.

No examples provided.

memory_write ~229

Creates a MemoryItem and optionally embeds it for semantic search. Contradiction detection is automatic — if new content conflicts with an existing memory of the same type/title, the old one is superseded. Use type='auto' to let the LLM decide the best category.

NameTypeReqDescription
agent_idstring
auto_classifyboolean
change_agentstring
contentstringyes
conversation_idstring
databasestring
embedboolean
embed_textstring
importancenumber
metadatastring
model_idstring
refresh_onstring
refresh_reasonstring
scopestring
sourcestring
timeoutnumber
titlestring
typestringyes
user_idstring
valid_fromstring
valid_tostring
variantstring

No output schema declared.

No examples provided.

tools_list_domains ~68

List m3 tool domains (memory, chatlog, files, entity, agent, tasks, conversations, diagnostics, admin) and their tool counts. Call `tools_load_domain` to expose a domain's full tool surface.

NameTypeReqDescription
databasestring
timeoutnumber

No output schema declared.

No examples provided.

tools_load_domain ~94

Register a tool domain's full surface for the current MCP session. Use when you need tools beyond the essentials (memory_search, memory_write, memory_get, chatlog_search, chatlog_write, files_search). Valid domains: memory, chatlog, files, entity, agent, tasks, conversations, diagnostics, admin.

NameTypeReqDescription
databasestring
domainstringyes
timeoutnumber

No output schema declared.

No examples provided.

Common questions

What is the M3 Memory MCP server?

M3 Memory is an MCP server listed in the public MCP registry as io.github.skynetcmd/m3-memory. Local-first memory, 100+ tools, 99.2% LongMemEval-S retrieval@10, hybrid search, GDPR, no cloud. This page covers its PyPI package (m3-memory).

Is the M3 Memory MCP server safe to use?

M3 Memory scores 53 out of 100 on VerifyMCP. We found no known CVEs affecting it as of 22 September 2026. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.

What tools does the M3 Memory MCP server expose?

M3 Memory exposes 10 tools: tools_list_domains, tools_load_domain, m3_call, memory_write, memory_supersede, and 5 more. Their descriptions and schemas cost roughly 1,467 tokens of context every time the server is loaded.

Is the M3 Memory MCP server still maintained?

M3 Memory is still listed as active in the MCP registry. We last reached this channel on 22 September 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.