AmberClaw — AI Memory Assistant Framework

A modular personal AI assistant framework that retains user preferences, contextual facts, and entity relationships across conversations.

Conversational LLMs typically start with zero memory every session, forcing users to re-explain project constraints, preferences, and workflows.

Persistent Memory Extraction: Implemented graph and vector memory storage using Mem0 to capture entity relationships and conversation facts.

Model Flexibility: Designed a model-agnostic layer supporting local models via Ollama and cloud LLM providers.

Extensible Tool System: Implemented deterministic tool routing and execution for messaging interfaces.

Lightweight Architecture: Built in Python with modular components for fast startup and low overhead.

GitHub: https://github.com/krishujeniya/AmberClaw