MARM Memory — Watchlist

MCP100 Assessment

Watchlist / Tracking

Why we’re tracking it

Track this specific mcp_server implementation and its adoption, setup drift, privilege model, and source activity.

Reviewed Oct 9, 2026 · Documentation review; complete setup workflow not independently tested

Best fit

Private persistent memory and code graphs for AI agents

What we know

Local-first memory MCP using SQLite, session context, codebase indexes and concept graphs with HTTP and stdio options.

Maintenance / activity evidence

Exact GitHub repository pushed_at 2026-10-08T01:26:35Z

Tracking evidence observed Oct 9, 2026. These dated signals do not establish current maintenance. Reviewed enrichment confidence: high.

Verification gaps / watch-outs

Installation and comprehensive security have not been independently audited.
Write/update stored memory and local indexes Reported action classes: Write/update stored memory and local indexes. Complete write authority is not verified.
Keyless loopback Python mode; bearer token when exposed or Docker
Sensitive access: Cross-session memory, Code snippets, Private notes

Set access boundaries on shared or exposed memory service
Persistent memory may contain secrets/private history and allow mutation; do not assume shared hosted access is scoped.
Local persistent memory saves user or repository context, credentials redaction is best effort; storage and sharing need care.
Unknown vulnerabilities and client-side tool approvals were not exhaustively evaluated.

Full authority, access and evidence limits are available in Review context and Access & risk.

Popularity

419 repository stars

Stars observed Oct 9, 2026.

Repository total for Lyellr88/marm-memory.

All limitations / Review context

What changed in documented permissions, distribution and MCP integration of MARM Memory before the next review?

No verified blocker to PUBLIC RADAR TRACKING within the bounded first-party evidence reviewed; material permissions and mitigations are explicitly disclosed.

Set access boundaries on shared or exposed memory service
Persistent memory may contain secrets/private history and allow mutation; do not assume shared hosted access is scoped.
Local persistent memory saves user or repository context, credentials redaction is best effort; storage and sharing need care.
Write/update stored memory and local indexes
Keyless loopback Python mode; bearer token when exposed or Docker
Depends on configured local process, connected account and remote APIs; review first-party privacy terms for any hosted component.
Reviewed write actions: Write/update stored memory and local indexes
Sensitive access: Cross-session memory, Code snippets, Private notes
Complete write authority is not verified.
Installation, client compatibility and comprehensive security have not been independently audited. Authentication and access depend on the configured services.
Unknown vulnerabilities and client-side tool approvals were not exhaustively evaluated.
Exact dated attributable activity evidence
Dated attributable implementation activity/usage not verified
production authorization and network isolation depend on deployment
Write actions/tool inventory not confirmed against exact tool documentation; listed action classes remain illustrative or context-dependent. Empty inventory does not establish read-only.
Shared memory requires isolation
Docker/exposed HTTP should use bearer key
Untrusted stored context may influence agents
No independent release lifecycle audit; do not claim precise activity date.

Original activity observations

Exact GitHub repository pushed_at 2026-10-08T01:26:35Z

Technical identity

github.com/Lyellr88/marm-memory

Exact implementation source: https://github.com/Lyellr88/marm-memory ↗
Reviewed resource type: MCP Server
Repository identity: Lyellr88/marm-memory
Package identity: pypi:marm-mcp-server

Evidence & Sources

Evidence basis: Recorded source evidence
Setup runtime verification: Not independently runtime-tested by MCP100 for this setup workflow.

Reviewed Oct 9, 2026 · Latest source observation Oct 9, 2026

GitHub repository ↗
lyellr88/marm-memory
Recorded repository material and project history.

Setup · Watchlist

Install & Agents

Publisher instructions.

Source evidence observed Oct 12, 2026.

Additional setup details required — see the specific gaps below.

Install / connect
Publisher instructions
Use the publisher-documented command or connection method below; complete the prerequisites and client configuration before expecting a usable MCP connection.
pip install marm-mcp-server marm-memory init --g-claude --g-codex --g-antigravity
Configure / authenticate
Publisher instructions
Install the marm-mcp-server package and initialize the intended agent integration. Start a separate managed HTTP runtime only if needed using marm-memory start, or use publisher stdio mode for manually wired clients; do not conflate the skill installer with an MCP process.
Check connection
MCP100 setup guide
Use marm-memory status for runtime, database, queue and graph status. marm-memory doctor provides publisher-supported installation diagnostics; MCP tool listing still needs inspection in the selected client.
marm-memory status
marm-memory doctor
Remove / uninstall / revoke
MCP100 setup guide
Use `marm-memory stop` to stop the managed MARM HTTP runtime; then remove its MCP registration/Skill from the chosen client if no longer needed. Do not delete `~/.marm` memory data as routine disconnect.
Advanced configuration · stop runtime
stop runtime · Review before use
marm-memory stop
python -m pip uninstall marm-mcp-server

Official Documentation & Sources

What it does

Local-first memory MCP using SQLite, session context, codebase indexes and concept graphs with HTTP and stdio options.

Capability: Files & Knowledge
Resource type: MCP Server
Best fit: Private persistent memory and code graphs for AI agents

Representative uses

Store/retrieve agent memories
Index codebase
Graph concept links
Share context across agents

Limitations

Shared memory requires isolation
Docker/exposed HTTP should use bearer key
Untrusted stored context may influence agents
Set access boundaries on shared or exposed memory service
No independent release lifecycle audit; do not claim precise activity date.

Reviewed access

Access & risk

Review the exact access boundary before connecting this MCP to accounts, credentials, or sensitive data.

Authentication: Keyless loopback Python mode; bearer token when exposed or Docker
Permissions: Write/update stored memory and local indexes
Data handling: Depends on configured local process, connected account and remote APIs; review first-party privacy terms for any hosted component.
Cost: unknown
Risk and cautions: Set access boundaries on shared or exposed memory service

Access scope and cautions

Write actions: Not verified Write/update stored memory and local indexes
Sensitive access: Cross-session memory Code snippets Private notes
Risk level: high
Security notes: Persistent memory may contain secrets/private history and allow mutation; do not assume shared hosted access is scoped. Local persistent memory saves user or repository context, credentials redaction is best effort; storage and sharing need care.

The reviewed evidence does not establish the complete write authority of this MCP. Consult the unresolved evidence before granting write access.

FAQ

Based on reviewed evidence as of 2026-10-09. Watchlist tracking is not a recommendation or a security certification.

Which tasks are documented for this implementation?

The reviewed tasks include: Store/retrieve agent memories; Index codebase; Graph concept links; Share context across agents.

What setup and authentication are documented?

Recorded installation/entrypoint step: pip install marm-mcp-server && marm-memory init. Recorded transport: stdio, streamable_http. Recorded environment-variable notes: MARM_API_KEY. No configuration values are supplied here. Prerequisites: Compatible MCP client. Client notes: Source docs and package may evolve; install according to confirmed release. These are the documented setup details, not a claim that installation or client compatibility was independently tested. Authentication: Keyless loopback Python mode; bearer token when exposed or Docker A demo or keyless description is not independently verified read-only authority. Check the documented tool permissions rather than inferring authority from that label.

Can it modify data or perform write operations?

Not verified. The available evidence does not establish the complete write authority. The partial inventory lists: Write/update stored memory and local indexes. Recorded permission context: Write/update stored memory and local indexes

What data-handling boundary should I consider?

Depends on configured local process, connected account and remote APIs; review first-party privacy terms for any hosted component. Recorded sensitive-access context: Cross-session memory; Code snippets; Private notes.

What limitations and risks should I check before use?

Set access boundaries on shared or exposed memory service. Shared memory requires isolation. Docker/exposed HTTP should use bearer key. Untrusted stored context may influence agents. No independent release lifecycle audit; do not claim precise activity date. Persistent memory may contain secrets/private history and allow mutation; do not assume shared hosted access is scoped. Local persistent memory saves user or repository context, credentials redaction is best effort; storage and sharing need care. Unknown vulnerabilities and client-side tool approvals were not exhaustively evaluated.