Forks may experiment, but official updates will always come from this repo.

🚀 Project Update: Q2 2026

I am currently finalizing the launch of SysDX, a forensic-grade Windows hardware diagnostics toolkit designed to catch intermittent PCIe flapping and TDR crashes through 800+ automated Pester tests. This production-heavy development phase has served as the ultimate stress test for MARM’s core logic, and I am now porting those real-world stability and memory optimizations back into this repository.

🛠️ Q2 Roadmap

• Session Persistence Fix: Resolving the marm_log_session state logic to ensure active sessions persist correctly before log entries fire.
• Intelligent Rate Limiting: Reducing the 10-minute localhost lockout and implementing a Retry-After header for smoother MCP client handling.
• Directory-Based Memory: Transitioning to a per-project SQLite architecture to allow isolated memory pools for different coding environments.
• Context Bloat Reduction: Implementing lazy-loading for documentation to prevent unnecessary token consumption during session startup.
• Automation Checkpoints: Adding an opt-in "Auto-Log" feature that triggers semantic summaries every N messages.
• Core Cleanup: Removing duplicate root files to establish a single source of truth and prevent future installation drift.

Thank you for your patience and support.

MARM Demo Video: Docker Install + Seeing Persistent AI Memory in Action

https://github.com/user-attachments/assets/c7c6a162-5408-4eda-a461-610b7e713dfe

This demo video walks through a Docker pull of MARM MCP and connecting it to Claude using the claude add mcp transport command and then shows multiple AI agents (Claude, Gemini, Qwen) instantly sharing logs and notebook entries via MARM’s persistent, universal memory proving seamless cross-agent recall and “absolute truth” notebooks in action.

Why MARM MCP: The Problem & Solution

Your AI forgets everything. MARM MCP doesn't.

Modern LLMs lose context over time, repeat prior ideas, and drift off requirements. MARM MCP solves this with a unified, persistent, MCP‑native memory layer that sits beneath any AI client you use. It blends semantic search, structured session logs, reusable notebooks, and smart summaries so your agents can remember, reference, and build on prior work—consistently, across sessions, and across tools.

MCP in One Sentence: MARM MCP provides persistent memory and structured session context beneath any AI tool, so your agents learn, remember, and collaborate across all your workflows.

The Problem → The MARM Solution

• Problem: Conversations reset; decisions get lost; work scatters across multiple AI tools.
• Solution: A universal, persistent memory layer that captures and classifies the important bits (decisions, configs, code, rationale), then recalls them by meaning—not keywords.

Before vs After

• Without MARM: lost context, repeated suggestions, drifting scope, "start from scratch."
• With MARM: session memory, cross-session continuity, concrete recall of decisions, and faster, more accurate delivery.

What MARM MCP Delivers

| Memory | Multi-AI | Architecture | |------------|--------------|------------------| | Semantic Search - Find by meaning using AI embeddings | Unified Memory Layer - Works with Claude, Qwen, Gemini, MCP clients | 18 Complete MCP Tools - Full Model Context Protocol coverage | | Auto-Classification - Content categorized (code, project, book, general) | Cross-Platform Intelligence - Different AIs learn from shared knowledge | Database Optimization - SQLite with WAL mode and connection pooling | | Persistent Cross-Session Memory - Memories survive across agent conversations | User-Controlled Memory - "Bring Your Own History," granular control | Rate Limiting - IP-based tiers for stability | | Smart Recall - Vector similarity search with context-aware fallbacks | | MCP Compliance - Response size management for predictable performance | | | | Docker Ready - Containerized deployment with health/readiness checks |

What Users Are Saying

“MARM successfully handles our industrial automation workflows in production. We've validated session management, persistent logging, and smart recall across container restarts in our Windows 11 + Docker environment. The system reliably tracks complex technical decisions and maintains data integrity through deployment cycles.”
@Ophy21, GitHub user (Industrial Automation Engineer)

“MARM proved exceptionally valuable for DevOps and complex Docker projects. It maintained 100% memory accuracy, preserved context on 46 services and network configurations, and enabled standards-compliant Python/Terraform work. Semantic search and automated session logs made solving async and infrastructure issues far easier. Value Rating: 9.5/10 - indispensable for enterprise-grade memory, technical standards, and long-session code management.” @joe_nyc, Discord user (DevOps/Infrastructure Engineer)

MARM MCP Server Guide

Now that you understand the ecosystem, here's info and how to use the MCP server with your AI agents

🚀 Quick Start for MCP (HTTP & Stdio)

Docker Install:

docker pull lyellr88/marm-mcp-server:latest
docker run -d --name marm-mcp-server -p 8001:8001 -v ~/.marm:/home/marm/.marm lyellr88/marm-mcp-server:latest
claude mcp add --transport http marm-memory http://localhost:8001/mcp

Local http Install:

pip install marm-mcp-server==2.2.6
pip install -r marm-mcp-server/requirements.txt
python marm-mcp-server
claude mcp add --transport http marm-memory http://localhost:8001/mcp

Stdio Install:

pip install marm-mcp-server==2.2.6
pip install -r marm-mcp-server/requirements_stdio.txt
<platform> mcp add --transport stdio marm-memory-stdio python "your/file/path/to/marm-mcp-server/server_stdio.py"
python marm-mcp-server/server_stdio.py

Docker (Fastest - 30 seconds):

docker pull lyellr88/marm-mcp-server:latest
docker run -d --name marm-mcp-server -p 8001:8001 -v ~/.marm:/home/marm/.marm lyellr88/marm-mcp-server:latest
claude mcp add --transport http marm-memory http://localhost:8001/mcp

Quick Local http Install:

pip install marm-mcp-server==2.2.6
pip install -r marm-mcp-server/requirements.txt
python marm-mcp-server
claude mcp add --transport http marm-memory http://localhost:8001/mcp

Http Manual JSON Configuration:

{
  "mcpServers": {
    "marm-memory": {
      "httpUrl": "http://localhost:8001/mcp",
      "authentication": {
        "type": "oauth",
        "clientId": "local_client_b6f3a01e",
        "clientSecret": "local_secret_ad6703cd2b4243ab",
        "authorizationUrl": "http://localhost:8001/oauth/authorize",
        "tokenUrl": "http://localhost:8001/oauth/token"
      }
    }
  }
}

Local Development Authentication (Development Only)

MARM includes mock OAuth 2.0 credentials for local testing—not a production authentication system.

Why hardcoded credentials? When developing locally, you don't have external OAuth providers (GitHub, Google, etc.). MARM includes dev credentials so you can test the full MCP authentication flow without external dependencies.

For local development, use these credentials:

• Client ID: local_client_b6f3a01e
• Client Secret: local_secret_ad6703cd2b4243ab

The server validates against these hardcoded values only during development.

For production deployment: Replace this entire section with real OAuth 2.1 authentication. These hardcod