Persistent local memory for Claude, Cursor & Codex

Persistent Local Memory for Claude, Cursor & Codex

This open-source MCP tool provides a robust solution for implementing persistent local memory for Large Language Models (LLMs) like Claude, Cursor, and Codex. Designed for developers, it leverages SQLite for efficient data storage and a knowledge graph for structured information retrieval. The key advantage is its entirely local operation, eliminating the need for cloud services or API keys, ensuring data privacy and control.

What it Does

This tool acts as an intermediary, enabling your LLM applications to retain and recall information across sessions. Instead of relying on ephemeral context windows, it stores relevant data locally in a structured format. This allows LLMs to build upon past interactions, access a growing knowledge base, and provide more consistent and contextually aware responses over time. The underlying SQLite database and knowledge graph facilitate efficient querying and retrieval of this persistent memory.

Key Features

• Persistent Local Storage: All memory is stored on your local machine using SQLite.
• Knowledge Graph Integration: Organizes information for more intelligent recall and reasoning.
• No Cloud Dependency: Operates entirely offline, enhancing privacy and security.
• API Key Free: Eliminates the need for external API keys for memory management.
• Open-Source: Full transparency and ability to customize or contribute.
• LLM Compatibility: Designed to integrate with Claude, Cursor, and Codex.

Who it's For

This tool is ideal for AI developers building applications that require long-term memory and context. This includes, but is not limited to:

• Chatbots and Virtual Assistants: For more personalized and consistent user experiences.
• Code Generation Tools: To maintain project context and coding history.
• Research and Analysis Agents: To build and query extensive knowledge bases.
• Personalized AI Companions: For applications that learn and adapt to individual users.
• Developers prioritizing data privacy and local control.