PluginBench
MCP Server
Active
Apache-2.0

io.github.Lyellr88/marm-mcp-server MCP Server

io.github.Lyellr88/marm-mcp-server

Local-first persistent memory layer for AI agents with semantic search and multi-agent support.

RepositoryRegistry
How to choose

What this kind of tool is for

  • Hand your agent a structured way to call a model or an inference API.
  • Compare outputs from two different providers on the same prompt.
  • Wire retrieval, embeddings, or generation into an existing agent workflow.

How to compare servers like this

Provider coverage:
Match the tool's supported providers and models to the ones your project uses.
Cost exposure:
Model-backed tools spend real API budget; check how usage and limits are configured.
Maintenance signal:
The provider landscape shifts weekly; prefer tools updated within the last few months.
Config surface:
Prefer tools whose environment-variable or config keys are documented plainly.

This page carries your copy-paste config: pick your MCP client in the install section on this page and use the copy button.

What is the io.github.Lyellr88/marm-mcp-server MCP server?

MARM MCP Server is a local memory infrastructure for AI agents like Claude and Cursor that provides persistent storage, semantic search, and session management. It exposes 7 focused MCP tools for logging, recall, notebooks, and memory maintenance, with built-in deduplication, hybrid search, and write-queue safety to keep token spend predictable as memory scales.

MARM gives AI agents a persistent local memory system so they can store decisions, retrieve context, reuse notebooks, and maintain long-running work without forgetting. It handles session routing, semantic search with FTS filtering and reranking, write-time consolidation, and supports solo use, multi-agent shared servers, and swarm deployments—all through a small 7-tool surface that keeps token costs low and predictable.

How to install io.github.Lyellr88/marm-mcp-server

Copy-paste configuration for popular MCP clients.

transport: stdio
Config generated by PluginBench — verify against the source before use.
~/Library/Application Support/Claude/claude_desktop_config.json
JSON config
{
  "mcpServers": {
    "marm-mcp-server": {
      "command": "uvx",
      "args": [
        "marm-mcp-server"
      ]
    }
  }
}

Tools & capabilities

Tools this server exposes to the agent.

  • marm_smart_recall — Hybrid recall with FTS5 filtering, semantic reranking, bounded fallback search, and chunk-aware scoring. Supports search_all=True, project/platform filters, and detail depth controls.
  • marm_log_entry — Add structured session log entries with automatic session/topic routing, summary-cache invalidation, and context summary preparation.
  • marm_log_show — Display all log entries and sessions, filterable by session or topic.
  • marm_delete — Delete a log session, log entry, or notebook entry by type.
  • marm_summary — Generate cached session summaries with intelligent truncation for LLM conversations.
  • marm_notebook — Unified notebook tool for add, use, show, status, or clear actions on notebook entries.
  • marm_compaction — Agent-assisted memory cleanup with status, candidates, review, stage, apply, and discard actions.

Use cases

  • Store and retrieve project decisions, code snippets, and configuration keys across multiple conversations without losing context
  • Share persistent memory across multiple AI agents working on the same project with write-queue safety and rate limiting
  • Search through long project histories using hybrid full-text and semantic search to find relevant past discussions and decisions
  • Maintain notebooks of reusable code, patterns, and workflows that persist across sessions and agents
  • Automatically deduplicate and compact memory over time to keep recall fast and token costs predictable as memory grows

io.github.Lyellr88/marm-mcp-server MCP server FAQ

What is MARM MCP Server?

MARM is a local-first persistent memory layer for AI agents. It stores session logs, notebooks, and semantic memories so Claude, Cursor, Gemini, and other MCP clients can retrieve context and decisions across conversations without forgetting.

Is MARM free?

Yes. MARM is open-source (Apache 2.0) and free to install via pip or Docker. Local STDIO mode requires no API key; HTTP mode requires a locally-generated key for security.

How do I install MARM in Cursor or Claude?

Install via pip (pip install marm-mcp-server), start the server (python -m marm_mcp_server), then add it to your client with: mcp add --transport http marm-memory http://localhost:8001/mcp. See the README for client-specific setup guides (Claude Code, Cursor, VS Code, Codex, Gemini, Qwen).

Does MARM require authentication?

Local STDIO mode (private single-agent) requires no key. HTTP mode (shared servers) requires a locally-generated API key for security. Docker HTTP deployments use the same key model.

Can multiple agents share one MARM server?

Yes. Use --swarm (200 RPM) or --swarm-max (600 RPM) flags when starting the HTTP server to enable multi-agent shared memory with write-queue safety and rate limiting.

What deployment options does MARM support?

Pip (local HTTP or STDIO), Docker (HTTP with key or STDIO), and swarm modes (--swarm, --swarm-max, --trusted). STDIO is best for private single-agent use; HTTP is best for shared multi-agent servers.

README (reference)

Source of truth, from the repository.

<div align="center"> <picture> <img src="https://raw.githubusercontent.com/Lyellr88/MARM-Systems/MARM-main/assets/marm-logo.svg" width="700" height="400"> </picture> <h1 align="center">MARM: Local-First Persistent Multi-Agent Memory Layer for MCP Clients v2.15.0</h1>

License Python FastAPI Docker Pulls PyPI Downloads PyPI Version MCP Registry

Discord Publish CodeQL MARM-Systems MCP server

Contributions welcome! Browse open issues to contribute, or join the MARM Discord to share workflows, get setup help, and connect with other builders.

</div>

Table of Contents

Why MARM MCP: The Problem & Solution

Your AI forgets everything. MARM MCP doesn't.

MARM MCP is a local memory infrastructure layer for AI agents. It gives Claude, Codex, Gemini, Qwen, IDE agents, and other MCP clients one persistent place to store decisions, retrieve context, reuse notebooks, and keep long-running work from drifting.

The point is not "more tools." MARM exposes 7 focused MCP tools and moves the heavy work behind the server: session routing, protocol delivery, hybrid recall, serialized writes, rate-limit presets, write-time consolidation, and agent-assisted compaction. Because the tool surface stays small, re-ranking filters results before they reach the model, and consolidation catches duplicates at write time, token spend stays low and predictable as workloads grow.

How It Works

LayerWhat it doesWhy it matters
Memory modelSessions, structured logs, notebooks, summaries, and semantic memoriesKeeps project history searchable instead of trapped in one chat
Scale layerSQLite WAL mode, connection pooling, serialized write queue, and HTTP rate-limit presetsLets one server support solo use, multi-agent work, and swarm-style bursts
Intelligence layerFTS filter, semantic re-rank, bounded semantic fallback, auto-classification, write-time consolidation, and compaction candidatesKeeps recall useful as memory grows instead of letting duplicates pile up
Token layerLightweight 7-tool surface, semantic re-rank before retrieval, and write-time deduplicationReduces tokens sent to the model on every recall and cost stays predictable as memory scales
Deployment layerPip, Docker, STDIO, HTTP, --swarm, --swarm-max, and --trustedLets you run private local memory or shared multi-agent memory with the same MCP surface

See Performance & Scaling Benchmarks for retrieval latency, concurrency, and write-cost numbers.

MARM Demo

https://github.com/user-attachments/assets/dabfe44f-689d-404f-a2c7-dcf8fa4ef0c1

MARM gives AI agents persistent local memory, shared context, write-queue safety, swarm presets, and hybrid recall so commands, config keys, and project meaning all stay reachable.

Start Now

Recommended: guided setup with marm-init

The easiest way to install MARM is to let your agent do the setup with you. marm-init turns the usual MCP setup mess into one guided conversation: Python or Docker, HTTP or STDIO, local or remote server, API keys, config paths, dashboard startup, and multi-agent linking for Claude, Codex, Gemini, Qwen, Cursor, VS Code, and other MCP clients. No hunting through install docs, no guessing which config file your client uses, and no rewriting the same connection by hand for every agent.

npx degit Lyellr88/MARM-Systems/skills

Then tell your agent: "Use the marm-init skill to set up MARM."

Manual pip install

pip install marm-mcp-server
If you are...Start the serverConnect your MCP client
Solo developer / researcherpython -m marm_mcp_server"agent" mcp add --transport http marm-memory http://localhost:8001/mcp
Private local STDIO usermarm-mcp-stdio"agent" mcp add --transport stdio marm-memory-stdio marm-mcp-stdio
Multiple agents sharing memorypython -m marm_mcp_server --swarm"agent" mcp add --transport http marm-memory http://localhost:8001/mcp
Private high-throughput swarmpython -m marm_mcp_server --swarm-max"agent" mcp add --transport http marm-memory http://localhost:8001/mcp
Trusted private lab/serverpython -m marm_mcp_server --trusted"agent" mcp add --transport http marm-memory http://localhost:8001/mcp

🚀 Quick Start for MCP (HTTP & STDIO)

Use this quick rule of thumb to choose your setup

  • Local HTTP/STDIO = fastest single-machine setup.
  • Docker HTTP = shared/always-on server (key required).
  • Docker STDIO = private containerized local use (no HTTP key).

Swarm / multi-agent note: The write queue is enabled by default to serialize memory writes through one worker. For shared HTTP deployments, use --swarm (200 RPM) or --swarm-max (600 RPM) when starting the server. --trusted disables rate limiting entirely for private deployments. STDIO is still best for private single-agent/local use. See MCP-HANDBOOK.md for more info.

<details> <summary><strong>Local pip HTTP (zero config)</strong></summary>

Local pip HTTP (zero config)

"agent" refers to claude, gemini, grok, qwen, or any MCP client. Codex uses --url instead of --transport to add MCP tools.

pip install marm-mcp-server
python -m marm_mcp_server
# Stuck on client setup? Open a Q&A thread: https://github.com/Lyellr88/MARM-Systems/discussions
# most agents use this --transport command
"agent" mcp add --transport http marm-memory http://localhost:8001/mcp
codex mcp add marm-memory --url http://localhost:8001/mcp

</details>

<details>
<summary><strong>Local pip STDIO</strong></summary>

#### Local pip STDIO

```bash
pip install marm-mcp-server
python -m marm_mcp_server.server_stdio
# most agents use this --transport command
"agent" mcp add --transport stdio marm-memory-stdio marm-mcp-stdio
codex mcp add marm-memory-stdio -- marm-mcp-stdio
</details> <details> <summary><strong>Local Python swarm modes (HTTP & STDIO)</strong></summary>

Local Python swarm modes (HTTP & STDIO)

Use HTTP when multiple agents need to share one live MARM server. STDIO is still best for private single-agent use because each client owns its own local process.

# HTTP shared server, normal multi-agent use
python -m marm_mcp_server --swarm

# HTTP shared server, heavier private swarm
python -m marm_mcp_server --swarm-max

# HTTP trusted private lab/server, rate limiting disabled
python -m marm_mcp_server --trusted

# STDIO remains keyless/private and does not use swarm flags
marm-mcp-stdio
</details>
<details> <summary><strong>Docker HTTP (key required)</strong></summary>

Docker HTTP (key required)

Docker HTTP requires an API key because it exposes MARM as a network server; STDIO stays local to the client process and does not need one.

# Step 1: generate key (do not add < > around the key)
docker run --rm lyellr88/marm-mcp-server:latest --generate-key

# Step 2: run server
docker pull lyellr88/marm-mcp-server:latest
docker run -d --name marm-mcp-server \
  -p 127.0.0.1:8001:8001 \
  -e SERVER_HOST=0.0.0.0 \
  -e MARM_API_KEY=your-generated-key \
  -v ~/.marm:/home/marm/.marm \
  lyellr88/marm-mcp-server:latest

# Step 3: connect client
"agent" mcp add --transport http marm-memory http://localhost:8001/mcp --header "Authorization: Bearer your-generated-key"
codex mcp add marm-memory --url http://localhost:8001/mcp --bearer-token-env-var MARM_API_KEY
</details> <details> <summary><strong>Docker HTTP swarm mode</strong></summary>

Docker HTTP swarm mode

# --swarm: write queue on, 200 RPM - recommended for multi-agent shared servers
docker run -d --name marm-mcp-server \
  -p 127.0.0.1:8001:8001 \
  -e SERVER_HOST=0.0.0.0 \
  -e MARM_API_KEY=your-generated-key \
  -v ~/.marm:/home/marm/.marm \
  lyellr88/marm-mcp-server:latest --swarm
</details> <details> <summary><strong>Docker STDIO (no HTTP key)</strong></summary>

Docker STDIO (no HTTP key)

docker run --rm -i \
  -v ~/.marm:/home/marm/.marm \
  --entrypoint python \
  lyellr88/marm-mcp-server:latest \
  -m marm_mcp_server.server_stdio
</details>
<details> <summary><strong>Support notes</strong></summary>

Most useful support info:

  • Docker HTTP requires a key; Docker STDIO does not.
  • If you get 401, verify key match and client restart after env var changes.
  • For full key setup, rotation, and troubleshooting: INSTALL-DOCKER.md
</details> <details> <summary><strong>Connect your client fast</strong></summary>

Connect Your Client Fast

Claude Code remains the recommended first setup path, but MARM also works with other MCP clients and IDE agents.

CLI clients - Claude Code · Codex · Gemini CLI · Qwen CLI · Linux variants · Docker/key

IDE agents - VS Code / Copilot Agent · Cursor · Docker/key IDE setup

Remote/API platforms - xAI / Grok Remote MCP · Platform integration

Using a client that isn't listed? Open an issue and let us know; client adapters are a first-class feature request.

</details>

MARM Dashboard

<div align="center"> <picture> <img src="https://raw.githubusercontent.com/Lyellr88/MARM-Systems/MARM-main/assets/marm-dashboard.png" width="700" height="400" </picture> </div>

A local web UI for browsing and managing your MARM memory; separate from the MCP server, reads and writes the same ~/.marm/marm_memory.db.

What it gives youHow it works
Browse/search/edit all memoriesDirect SQLite, no MCP required
Manage sessions and protocol logsRuns on port :8002 alongside MCP on :8001
Notebook CRUD with inline editorSame auth model (MARM_API_KEY) as the MCP server
Delete-all with count confirmationDocker image included; WAL mode handles concurrent access
View the write queue in real timePulls live data from the write queue
# Quick start (pip)
cd marm-dashboard
pip install -e .
python -m marm_dashboard --open
# Docker (same key and volume as MCP)
docker build -t marm-dashboard:local ./marm-dashboard
docker run --rm -p 127.0.0.1:8002:8002 \
  -e MARM_API_KEY=your-key \
  -v ~/.marm:/home/marm/.marm \
  marm-dashboard:local

See marm-dashboard/README.md for the full guide.

Complete MCP Tool Suite (7 Tools)

<div align="center"> <picture> <img src="https://raw.githubusercontent.com/Lyellr88/MARM-Systems/MARM-main/assets/mcp-tools.png" width="700" height="400" </picture> </div>

💡 Pro Tip: You don't need to manually call these tools! Just tell your AI agent what you want in natural language:

  • "Claude, log this session as 'Project Alpha' and add this conversation as 'database design discussion'"
  • "Remember this code snippet in your notebook for later"
  • "Search for what we discussed about authentication yesterday"

The AI agent will automatically use the appropriate tools. Manual tool access is available for power users who want direct control.

CategoryToolDescription
Memory Intelligencemarm_smart_recallHybrid recall with FTS5 filtering, semantic reranking, bounded fallback search, and chunk-aware scoring for long memories. Supports search_all=True, project/platform filters, and detail=1/2/3 depth controls
Logging Systemmarm_log_entryAdd structured session log entries. Session/topic routing, summary-cache invalidation, and context summary preparation are handled by the server
marm_log_showDisplay all entries and sessions (filterable)
marm_deleteDelete a log session, log entry, or notebook entry (type="log"|"notebook")
Reasoning & Workflowmarm_summaryGenerate cached session summaries with intelligent truncation for LLM conversations
Notebook Managementmarm_notebookUnified notebook tool: add, use, show, status, or clear entries with action="add"|"use"|"show"|"status"|"clear"
Memory Maintenancemarm_compactionUnified compaction workflow with action="status"|"candidates"|"review"|"stage"|"apply"|"discard" for agent-assisted memory cleanup

A Deeper Look

MARM keeps MCP discovery lean with 7 tools by grouping domain operations behind explicit parameters like marm_notebook(action=...), marm_delete(type=...), and marm_compaction(action=...). Behind those tools, the server handles lifecycle setup, protocol refresh, docs indexing, date context, summary-cache maintenance, write queue handling, project/platform attribution, and health checks.

Under the hood, MARM uses SQLite WAL mode, connection pooling, serialized writes, HTTP swarm presets, safe local defaults, FTS→semantic reranking, bounded fallback search, chunk-aware long-memory recall, and summary/context/full recall depths to keep memory fast, stable, and token-efficient as projects grow.

For a deeper look into the MCP behavior, tool parameters, automation, and workflows, see MCP-HANDBOOK.md and FAQ.md.

Performance & Scaling Benchmarks

<div align="center"> <picture> <img src="https://raw.githubusercontent.com/Lyellr88/MARM-Systems/MARM-main/assets/marm-bench.png" width="700" height="400" </picture> </div>

MARM is tuned for fast recall first, even as memory grows and multiple agents hit the same server.

1. Retrieval Latency Scaling

Session Size ($N$)Min LatencyMedian Latencyp95 Latency
N = 10012.0 ms17.4 ms20.8 ms
N = 50012.4 ms20.5 ms22.6 ms
N = 1,00015.9 ms23.3 ms25.1 ms
N = 4,00023.1 ms30.4 ms31.3 ms

2. Multi-Agent Concurrency

  • Parallel recall wins: 10 concurrent recalls completed in 316.3ms vs 647.0ms serial, a 51% time reduction.

3. Write-Time Ingestion Cost

  • Write-time tradeoff: consolidation raises median ingest from 20.3ms to 85.2ms (4.2x) so dedupe/clustering cost stays off the hot recall path.

Benchmarks used a real SQLite database and the live all-MiniLM-L6-v2 encoder on local hardware. Reproduce them: marm-mcp-server/scripts/bench_hotpath.py

⭐ Star the Project

If MARM helps with your AI memory needs, please star the repository to support development!

<div align="center"> <a href="https://star-history.com/#Lyellr88/MARM-Systems&Date"> <img src="https://api.star-history.com/svg?repos=Lyellr88/MARM-Systems&type=Date" width="700" height="400" alt="MARM Systems star history chart"> </a> </div>

Contributing

MARM welcomes contributors at every level. Code helps, but so do docs, setup notes, client testing, bug reports, benchmarks, and real workflow feedback from people using AI tools every day.

Good places to help:

  • Test MARM with more MCP clients, IDE agents, and operating systems
  • Improve docs, screenshots, examples, and platform-specific setup notes
  • Report bugs or confusing install steps with clear reproduction details
  • Share memory workflows, agent habits, and tool ideas from real use
  • Check out open issues

💡 Want to get your name on this list? Check out our CONTRIBUTING.md guide to get started!

Join the MARM Community

Help build the future of AI memory - no coding required!

Connect: MARM Discord | GitHub Discussions

License & Usage Notice

MARM is released under the Apache 2.0 License, and forks, experiments, and integrations are welcome. If you build on it, please make unofficial versions easy to distinguish from releases published by the official MARM repository so users know what they are installing.

Project Documentation

Usage Guides

  • MCP-HANDBOOK.md - Complete MCP server usage guide with commands, workflows, and examples
  • PROTOCOL.md - MCP operating protocol
  • FAQ.md - Answers to common questions about using MARM

MCP Server Installation

Project Information

Related MCP servers

Claude Code creature collection game. Hatch buddies, pet to boost roll luck, complete the BuddyDex.

0
TypeScript
MIT
View repository →

Get a second opinion from a different AI model on code reviews, architecture, and diffs.

18
TypeScript
MIT
View repository →

Bridge Claude with Gemini CLI for AI-to-AI collaboration, code review, and second opinions

18
TypeScript
MIT
View repository →

Consult Grok through xAI's metered API with exact model selection and no fallback.

18
TypeScript
MIT
View repository →

Bridge Claude with Google's Antigravity CLI for independent code review and second opinions from a different model.

18
TypeScript
MIT
View repository →

Get a second opinion on code from a different AI model—Claude, Codex, Grok, Gemini, Ollama, or Antigravity.

18
TypeScript
MIT
View repository →