AutoMem MCP: Give Your AI Perfect Memory

One command. Infinite memory. Perfect recall across all your AI tools.

npx @verygoodplugins/mcp-automem setup

Your AI assistant now remembers everything. Forever. Across every conversation.

Works with Claude Desktop, Cursor IDE, Claude Code, GitHub Copilot (coding agent), ChatGPT, ElevenLabs, OpenAI Codex, Google Antigravity - any MCP-compatible AI platform.

The Problem We Solve

Every AI conversation starts from zero. Claude forgets your coding style. Cursor can't learn your patterns. Your assistant doesn't remember yesterday's decisions.

Until now.

AutoMem MCP connects your AI to persistent memory powered by AutoMem - a graph-vector memory service.

What You Get

🧠 Persistent Memory Across Sessions

• AI remembers decisions, patterns, and context forever
• Works across all MCP platforms - Claude Desktop, Cursor, Claude Code, OpenAI Codex, Google Antigravity
• Cross-device sync - same memory on Mac, Windows, Linux

🏆 Graph-Vector Architecture

• 11 public authorable relationship types between memories (recall results may also include read-only system/internal relations that are not valid associate_memories inputs)
• Research-validated approach (HippoRAG 2: 7% better associative memory)
• Sub-second retrieval even with millions of memories

🚀 Works Everywhere You Code

See It In Action

Claude Desktop with Personal Preferences

Claude automatically recalls memories using the Personal Preferences template

Cursor IDE with Memory Rules

Cursor uses automem.mdc rule to automatically recall and store memories

Claude Code with Session Memory

Git commits, builds, and deployments automatically stored to memory

OpenAI Codex with Memory Rules

OpenAI Codex uses config.toml to automatically recall and store memories

Hermes Agent Recalling Memory

Hermes is a terminal agent, so AutoMem's per-turn recall is invisible by design — it's injected into the model payload before each turn and never printed. hermes automem debug-recall surfaces the exact block:

The <memory-context> block AutoMem injects ahead of every Hermes turn — normally never shown in the terminal. Captured against a synthetic demo dataset.

A live one-shot turn then answers from that recalled memory alone. The staging port exists only in the seeded memories, so a correct answer proves recall fired:

hermes -z recalls the seeded Project Nimbus memory and answers Port 7341 — concisely, honoring the recalled output preference.

Your AI Learns Your Code Style

// After 1 week, your AI writes EXACTLY like you
// ✅ It knows you prefer early returns
// ✅ It uses your specific variable naming
// ✅ It matches your comment style
// ✅ It follows YOUR patterns, not generic best practices

Decisions That Feel Like Yours

User: "Should we use Redis for this?"

Without AutoMem:
"Consider RabbitMQ, Kafka, or AWS SQS based on your needs..."

With AutoMem:
"Based on your pattern of preferring boring technology that works,
and your positive experience with Redis in Project X (March 2024),
yes. You specifically value operational simplicity over feature
richness - Redis fits perfectly."

Quick Start

1. Set Up AutoMem Service

You need a running AutoMem service (the memory backend). Choose one:

Option A: Local Development (fastest, free)

git clone https://github.com/verygoodplugins/automem.git
cd automem
make dev

Service runs at http://localhost:8001 - perfect for single-machine use.

Option B: Railway Cloud (recommended for production)

One-click deploy with $5 free credits. Typical cost: ~$0.50-1/month after trial.

👉 AutoMem Service Installation Guide - Complete setup instructions for local, Railway, Docker, and production deployments.

2. Install MCP Client

Claude Desktop - One-Click Install

Download and double-click to install AutoMem in Claude Desktop:

⬇️ Download AutoMem for Claude Desktop (.mcpb)

After installing:

1. Claude Desktop will prompt you for your AutoMem Endpoint (http://127.0.0.1:8001 for local)
2. Optionally enter your API Key (required for Railway, skip for local)
3. Click Enable

Then add the paste-ready Personal Preferences starter from templates/CLAUDE_DESKTOP_INSTRUCTIONS.md. That's it: Claude now has persistent memory and knows when to use it.

Other Platforms

Connect your AI tools to the AutoMem service you just started.

# Guided setup - creates .env and prints config for your AI platform
npx @verygoodplugins/mcp-automem setup

When prompted:

• AutoMem Endpoint: http://localhost:8001 (or your Railway URL if deployed)
• API Key: Leave blank for local development (or paste your token for Railway)

The wizard will:

• ✅ Save your endpoint and API key to .env
• ✅ Generate config snippets for Claude Desktop/Cursor/Code
• ✅ Validate connection to your AutoMem service

3. Platform-Specific Setup

For Claude Desktop:

# Setup prints config snippet - just paste into claude_desktop_config.json
npx @verygoodplugins/mcp-automem setup

Then paste templates/CLAUDE_DESKTOP_INSTRUCTIONS.md into Claude Desktop → Settings → Profile → Personal Preferences.

For Cursor IDE:

# Or use CLI to install automem.mdc rule file
npx @verygoodplugins/mcp-automem cursor

Note: After one-click install, configure your AUTOMEM_API_URL in ~/.cursor/mcp.json or Claude Desktop config

For Claude Code:

Option A: CLI Setup (Recommended)

# Installs SessionStart hook and MCP permissions
npx @verygoodplugins/mcp-automem claude-code

This is the supported Claude Code integration path.

On Windows, this compatibility path currently assumes a POSIX shell environment such as Git Bash, MSYS2, or WSL. bash, jq, and Python must be available. This is not full native Windows hook support yet.

Option B: Plugin (Deprecated)

# In Claude Code, install the plugin:
/plugin marketplace add verygoodplugins/mcp-automem
/plugin install automem@verygoodplugins-mcp-automem

The marketplace plugin is deprecated and kept only as a migration bridge for one release. Use npx @verygoodplugins/mcp-automem claude-code for new installs.

Migration details: DEPRECATION.md

For OpenAI Codex:

# Add to your Codex MCP configuration
npx @verygoodplugins/mcp-automem config --format=json

# Optional: add memory-first rules to this repo
npx @verygoodplugins/mcp-automem codex

For Hermes Agent (Nous Research):

# Registers `automem` MCP server under ~/.hermes/config.yaml
# and installs Hermes-specific AGENTS.md rules.
npx @verygoodplugins/mcp-automem hermes

# Optional: replace Hermes' built-in memory provider with AutoMem
npx @verygoodplugins/mcp-automem hermes --mode provider

# Advanced: ambient provider recall plus MCP write/recall tools
npx @verygoodplugins/mcp-automem hermes --mode both

For Google Antigravity:

1. Open the MCP Store from the ... menu at the top of the editor's agent panel
2. Click Manage MCP Servers and then View raw config
3. Paste the config from templates/antigravity/mcp_config.json into ~/.gemini/antigravity/mcp_config.json
4. Restart or reload Antigravity so the memory server is available

👉 Google Antigravity Setup for the full flow and verification steps

👉 Full Installation Guide for detailed MCP client and platform-specific setup

New: Remote MCP via HTTP

You can now connect AutoMem to platforms that support remote MCP via Streamable HTTP (recommended) or SSE transport via an optional sidecar service (deployable to Railway or any Docker host).

• ChatGPT (Developer Mode custom connectors)
• Claude.ai (web) and Claude Mobile (iOS/Android)
• ElevenLabs Agents Platform

Quick connect URLs (after deploying the sidecar):

• Streamable HTTP (recommended): https://<your-mcp-domain>/mcp?api_token=<AUTOMEM_API_TOKEN>
• SSE (legacy): https://<your-mcp-domain>/mcp/sse?api_token=<AUTOMEM_API_TOKEN>
• ElevenLabs: https://<your-mcp-domain>/mcp with header Authorization: Bearer <AUTOMEM_API_TOKEN>

See the Installation Guide for complete steps and deployment options.

Remote MCP Platforms in Action

ChatGPT Developer Mode: Add your MCP endpoint as a custom connector

ChatGPT using AutoMem memories via remote MCP

Claude.ai website connected to AutoMem via remote MCP

Claude Mobile (iOS) connected to AutoMem via remote MCP

What Happens Next

Architecture

┌─────────────────────────────────────────────┐
│         Your AI Platforms                   │
│  Claude Desktop │ Cursor │ Claude Code      │
└──────────────┬──────────────────────────────┘
               │ MCP Protocol
               ▼
┌──────────────────────────────────────────────┐
│   @verygoodplugins/mcp-automem (this repo)  │
│   • Translates MCP calls → AutoMem API      │
│   • Platform integrations & rules           │
│   • Handles authentication                   │
└──────────────┬───────────────────────────────┘
               │ HTTP API
               ▼
┌──────────────────────────────────────────────┐
│        AutoMem Service (separate repo)       │
│        github.com/verygoodplugins/automem    │
│   ┌────────────┐      ┌────────────┐        │
│   │  FalkorDB  │      │   Qdrant   │        │
│   │  (Graph)   │      │ (Vectors)  │        │
│   └────────────┘      └────────────┘        │
└──────────────────────────────────────────────┘

This repo (mcp-automem):

• MCP client that connects AI platforms to AutoMem
• Platform-specific integrations (Cursor rules, Claude Code hooks, etc.)
• Setup wizards and configuration tools

AutoMem service:

• Backend memory service with graph + vector storage
• Deployment guides (local, Railway, Docker, production)
• API server with FalkorDB + Qdrant

Features

Core Memory Operations

• store_memory — Save memories with content, tags, importance, metadata. Two modes:
  • Single (default): top-level content plus optional fields, including embedding, t_valid, t_invalid, custom id.
  • Batch: memories: [...] (≤500 items) for bulk ingestion. Per-item id/embedding/t_valid/t_invalid are not supported in batch mode.
• recall_memory — Three modes selected by which params you pass:
  • ID fetch: memory_id → fetches one memory by ID; updates last_accessed.
  • Tag enumeration: tags + exhaustive: true → paginated exact-match listing for cleanup/audit workflows where ranked recall undercounts. Pair with limit (≤200) and offset; returns has_more.
  • Ranked retrieval (default): hybrid search across vector, keyword, tags, recency, with optional graph expansion and exclude_tags to filter out unwanted scopes.
• associate_memories — Create relationships (11 public authorable types; recall results may also include read-only system relations)
• update_memory — Modify existing memories
• delete_memory — Two modes:
  • Single (default): memory_id → removes one memory and its embedding.
  • Bulk-by-tag: tags: [...] → bulk-delete all memories matching ANY tag (exact, case-insensitive). No dry-run; verify with recall_memory({ tags, exhaustive: true }) first.
• check_database_health — Monitor service status

Advanced Recall (v0.8.0+)

Multi-hop Reasoning - Answer complex questions like "What is Amanda's sister's career?"

mcp__memory__recall_memory({
  query: "What is Amanda's sister's career?",
  expand_entities: true, // Finds "Amanda's sister is Rachel" → memories about Rachel
});

Context-Aware Coding - Recall prioritizes language and style preferences

mcp__memory__recall_memory({
  query: "error handling patterns",
  language: "typescript",
  context_types: ["Style", "Pattern"],
});

Platform Integrations

Cursor IDE

• ✅ Memory-first rule file (automem.mdc in .cursor/rules/)
• ✅ Automatic memory recall at conversation start
• ✅ Auto-detects project context (package.json, git remote)
• ✅ Global user rules option for all projects
• ✅ Simple setup via CLI or one-click install

Claude Code

• ✅ MCP permissions for memory tools
• ✅ Memory rules in CLAUDE.md guide Claude's memory usage
• ✅ Simple setup - just permissions, Claude decides what to store

Claude Desktop

• ✅ Direct MCP integration
• ✅ Paste-ready Personal Preferences starter template
• ✅ Full memory API access

Why AutoMem MCP?

vs. Building Your Own

• ✅ 2 years of R&D already done
• ✅ Research-validated architecture (HippoRAG 2, MELODI, A-MEM)
• ✅ Working integrations across all MCP platforms
• ✅ Active development and community

vs. Other Memory Solutions

• ✅ True graph relationships (not just vector similarity)
• ✅ Universal MCP compatibility (works with any MCP client)
• ✅ 7 memory types (Decision/Pattern/Preference/Style/Habit/Insight/Context)
• ✅ Self-hostable ($5/month vs $150+ for alternatives)

vs. Native AI Memory

• ✅ Persistent across sessions (not just context window)
• ✅ Cross-platform (same memory in Claude, Cursor, Code)
• ✅ Structured relationships (not just RAG)
• ✅ Infinite scale (no context window limits)

Real-World Results

Code Review That Knows Your Standards

Before AutoMem:
"Consider adding error handling here."

After AutoMem:
"Missing your standard try/except pattern. Based on your PR#127
review comments, you always wrap database calls with specific
logging for timeouts. Apply the same pattern here?"

Decisions With Context

Before AutoMem:
"Both approaches have tradeoffs..."

After AutoMem:
"You chose PostgreSQL over MongoDB for similar use case in Q1 2024.
Your decision memo cited team expertise and operational simplicity.
Same factors apply here - go with Postgres."

Documentation

MCP Client & Integrations (this repo)

• 📦 Installation Guide - MCP client setup for all platforms
• 🌐 Remote MCP via SSE - Connect ChatGPT, Claude Web/Mobile, ElevenLabs
• 🎯 Cursor Setup - IDE integration with rules
• 🤖 Claude Code Setup - Memory rules integration
• ⚠️ Deprecations - Claude Code plugin migration and removal plan
• 🚀 OpenAI Codex Setup - Codex CLI/IDE/Cloud integration
• 🪐 Google Antigravity Setup - Raw MCP config via Antigravity's MCP Store
• 📖 MCP Tools Reference - All memory operations

AutoMem Service (separate repo)

• 🏗️ AutoMem Service - Backend repository
• 🚀 Service Installation - Local, Railway, Docker deployment
• ⚙️ API Documentation - REST API reference
• 🧪 Evaluation Lab - Exploratory recall-quality benchmarks and ruleset A/B testing

The Science Behind AutoMem

The AutoMem service implements cutting-edge 2025 research:

• HippoRAG 2 (OSU, June 2025): Graph-vector approach achieves 7% better associative memory
• A-MEM (July 2025): Dynamic memory organization with Zettelkasten principles
• MELODI (DeepMind, 2025): 8x memory compression without quality loss
• ReadAgent (DeepMind, 2024): 20x context extension through gist memories

This MCP package provides the bridge between your AI and that research-validated memory system.

Community & Support

• 💬 Discord - Join the community, get help, share feedback
• 🐦 X Community - Discussion and updates
• 📣 @automem_ai - Official announcements
• 📦 NPM Package - This MCP client
• 🔬 AutoMem Service - Backend repo with deployment guides
• 🐛 GitHub Issues - Bug reports and feature requests

Quick Links

MCP Client Setup

• Installation Guide - MCP client setup for all platforms
• Cursor Integration - IDE rules and configuration
• Claude Code Setup - Memory rules integration
• Deprecations - Claude Code plugin migration and removal plan
• OpenAI Codex - Codex integration
• Google Antigravity - Antigravity MCP setup
• Changelog - Release history

AutoMem Service

• Service Repository - Backend source code
• Service Installation - Local, Railway, Docker deployment

Contributing

We welcome contributions! Please:

1. Fork the repository
2. Create a feature branch
3. Make your changes with tests
4. Submit a pull request with a Conventional Commit title such as fix:, feat:, docs:, or chore:
5. Do not prefix the PR title with labels like [codex] or [wip] because the squash-merge commit is taken from the PR title

License

MIT - Because great memory should be free.

Ready to give your AI perfect memory?

npx @verygoodplugins/mcp-automem setup

Built with obsession. Validated by neuroscience. Powered by graph theory. Works with every MCP-enabled AI.

Designed by Jack Arturo at Very Good Plugins 🧡

Transform your AI from a tool into a teammate. Start now.