A-MEM
StaleDescription
An agentic memory system for LLM agents inspired by human memory mechanisms, enabling dynamic memory generation, retrieval, and consolidation with automatic memory evolution and self-organization.
Key Features
- Dynamic memory organization based on Zettelkasten principles for LLM agents
- ChromaDB-powered vector storage with semantic similarity search
- Automatic memory evolution that creates connections between related memories
- Structured note generation with tags, context, keywords, and timestamps
- Support for multiple LLM backends including OpenAI and local Ollama
- Full CRUD operations: add, read, update, delete, and search memories
Use Cases
💡 Long-term memory for LLM agents that need to retain and recall past interactions
💡 Research note management with automatic linking of related concepts
💡 Building knowledge bases that evolve and self-organize over time
💡 Enhancing conversational agents with persistent contextual awareness
💡 Academic and scientific memory systems for literature review and synthesis
Categories
Quick Start
Clone the repo, create a virtual env, and `pip install .`. Initialize the system with `AgenticMemorySystem(model_name='all-MiniLM-L6-v2', llm_backend='openai', llm_model='gpt-4o-mini')`. Add memories via `memory_system.add_note('content')`, search with `search_agentic('query', k=5)`, and update/delete as needed.