Simplemem

Efficient Lifelong Memory for LLM Agents - semantic compression, cross-session memory, and intent-aware retrieval

SimpleMem Skill

Integrates SimpleMem: Efficient Lifelong Memory for LLM Agents into OpenClaw.

What it does

SimpleMem provides semantic memory compression and retrieval for agents:

• Store: Compresses interactions into compact memory units
• Synthesize: Merges related memories on-the-fly
• Retrieve: Intent-aware planning for efficient context retrieval

Installation

# Install Python dependency
pip install simplemem

# Or via repo
git clone https://github.com/aiming-lab/SimpleMem.git
cd SimpleMem
pip install -r requirements.txt

Configuration (Optional - Full Features)

For full SimpleMem features, set your OpenAI API key:

$env:OPENAI_API_KEY = "your-openai-key"

Without API key: Uses JSON fallback (basic keyword search) With API key: Uses full SimpleMem with semantic embeddings

Usage

PowerShell Script

# Agregar memoria
.\simplemem.ps1 -Action add -Content "El usuario prefiere cafe con leche de avena"

# Buscar memorias
.\simplemem.ps1 -Action search -Query "cafe"

# Ver estadisticas
.\simplemem.ps1 -Action stats

Python API

from simplemem import SimpleMemSystem, set_config, SimpleMemConfig

# With API key (full features)
config = SimpleMemConfig()
config.openai_api_key = "your-key"
set_config(config)
system = SimpleMemSystem()

# Add memory
system.add("User preference: coffee with oat milk", user_id="user1")

# Retrieve
results = system.retrieve("What does user like?", user_id="user1")

Key Features

• Cross-session memory: Persistent across conversations (64% better than Claude-Mem)
• Semantic compression: 43.24% F1 on LoCoMo benchmark
• Fast retrieval: 388ms average retrieval time
• Multi-index: Semantic + Lexical + Symbolic layers
• Fallback: JSON-based storage when no API key available

Files

• simplemem.py - Main Python wrapper
• simplemem.ps1 - PowerShell CLI script
• data/ - Storage directory (created on first use)

Credits

• Repo: https://github.com/aiming-lab/SimpleMem
• Paper: https://arxiv.org/abs/2601.02553
• Discord: https://discord.gg/KA2zC32M