A-MEM is an agentic memory system designed to enhance LLM agents by providing dynamic organization and intelligent retrieval of historical experiences. It targets developers building sophisticated LLM agents that require more than basic memory storage, offering a Zettelkasten-inspired approach for interconnected knowledge networks.

How It Works

A-MEM leverages ChromaDB for efficient vector embedding storage and semantic similarity search. When new memories are added, the system generates structured notes with attributes, tags, and contextual descriptions. It then analyzes existing memories to establish meaningful links based on semantic similarity, enabling continuous memory evolution and refinement through agent-driven updates.

Quick Start & Requirements

• Install via pip install -r requirements.txt within a Python virtual environment (venv or Conda).
• Requires Python 3.9+.
• Supports OpenAI and Ollama LLM backends.
• Embedding model all-MiniLM-L6-v2 is used by default.
• Official documentation and usage examples are available in the README.

Highlighted Details

• Dynamic memory organization based on Zettelkasten principles.
• Intelligent indexing and linking of memories via ChromaDB.
• Comprehensive note generation with structured attributes.
• Interconnected knowledge networks and continuous memory evolution.
• Supports OpenAI (GPT-4, GPT-3.5) and Ollama LLM backends.

Maintenance & Community

The project is associated with agiresearch. The primary citation points to a 2025 arXiv preprint. No community links (Discord, Slack) are provided in the README.

Licensing & Compatibility

Licensed under the MIT License, permitting commercial use and integration with closed-source projects.

Limitations & Caveats

The repository is presented as a memory system to facilitate agent construction, with a separate repository linked for reproducing paper results. The README does not detail specific performance benchmarks or potential limitations of the memory evolution process.