C.O.R.E. (Contextual Observation & Recall Engine) provides a private, portable, and user-owned memory layer for Large Language Models (LLMs). It addresses the need for persistent, traceable context across different AI applications, enabling personalized and auditable interactions. The target audience includes developers and power users seeking to enhance LLM applications with dynamic, temporal knowledge graphs.

How It Works

C.O.R.E. functions as a temporal knowledge graph, storing facts as "Statements" with rich metadata including source, timestamp, and rationale. This contrasts with simpler memory systems by providing full transparency and auditability, allowing users to trace the origin and evolution of information. This dynamic approach facilitates complex queries about changes over time and individual knowledge provenance.

Quick Start & Requirements

• Local Setup: Requires Docker and an OpenAI API Key.
• Installation: Clone the repository, copy .env.example to .env, and run docker-compose up.
• Access: Navigate to http://localhost:3000 and use a Magic Link for login.
• Documentation: C.O.R.E. Documentation
• Demo: C.O.R.E Demo Video

Highlighted Details

• Temporal Knowledge Graph: Stores facts with history, source, and context.
• Auditability: Enables tracing information origins and changes.
• Integration: Supports connection with tools like Cursor via MCP server.
• API Access: Provides RESTful APIs for programmatic ingestion and search.

Maintenance & Community

• Community: Discord
• Status: Actively improving Llama model support; currently suboptimal.

Licensing & Compatibility

• License: Not explicitly stated in the README.
• Compatibility: Designed for integration with LLM tools; requires OpenAI API key for core functionality.

Limitations & Caveats

The project is in early stages, with Llama model support being actively developed and currently suboptimal. Features like user-controlled sharing, granular API permissions, and role-based access control are still in progress.