AuraCoreCF is a pioneering local-first desktop AI operating system designed for persistent interactions with artificial intelligence. Unlike traditional chat-based interfaces, AuraCoreCF emphasizes that language generation is merely one layer of a more complex cognitive framework. The system integrates crucial aspects such as memory, continuity, planning, identity management, safety protocols, and tool routing, all handled by its proprietary runtime before engaging AI models for responses.

Features

AuraCoreCF sets itself apart as a cognitive runtime, providing a multitude of capabilities:

• Persistent User Context: User profiles and sessions are stored locally and securely, enabling tailored interactions.
• Cognitive Field Runtime: Features active memory management, salience, stabilization, and reinforcement learning loops.
• Temporal Continuity: Supports handling of episodic memories and ensures users maintain a consistent experience over time.
• Sophisticated Memory Management: Incorporates working memory, fact extraction, self-modeling, and recall orchestration to enhance user interaction.
• Proposal-First Execution: Prioritizes user proposals for tool execution, incorporating permissions and maintaining audit trails.
• Local-First Desktop UX: Provides a user-friendly interface with settings, profile management, and robust billing features.
• Configurable Inference Routing: Offers flexibility to operate with either a local model (via Ollama) or cloud-based APIs.

Unique Experience

AuraCoreCF redefines cognitive interaction by distinguishing between cognition and expression. User messages undergo analysis, grounding in memory, and validation against safety and identity constraints before they are passed to the language model for rendering. This separation fosters a more cohesive and durable user experience across sessions.

Inference Modes

Users can choose from three inference modes based on their needs:

• Auto: Automatically utilizes a cloud model if available; defaults to local Ollama if cloud settings are absent.
• Local Ollama: Strictly uses a local Ollama server.
• Cloud API: Connects exclusively to a configured cloud provider for model inference.

Examples:

• Local Ollama Endpoint: http://127.0.0.1:11434
• OpenAI Cloud URL: https://api.openai.com/v1
• Anthropic Cloud URL: https://api.anthropic.com/v1

If model settings remain unconfigured, the application will utilize its local-first defaults.

Architecture Overview

AuraCoreCF is structured into multiple collaborative layers:

• Runtime: For boot processes, orchestration, and lifecycle management.
• Cognitive Layer: Manages cognitive processes ensuring coherence and continuity.
• Memory Layer: Responsible for memory handling and shaping.
• Interaction Layer: Focuses on response planning and payload management.
• Security Layer: Implements safety and threat analysis measures.
• Middleware and Tools: Handles permissions and execution routing.
• User Interface Layer: Encompasses the primary desktop shell and chat experience.
• Auth Layer: Manages user accounts and session data.

For detailed architectural insights, refer to the ARCHITECTURE.md file in the repository.

Product Surfaces

AuraCoreCF currently encompasses:

• Sign-in and account management systems
• A main desktop chat interface
• User preference and profile administration
• AI model configuration options
• Proposal and tool interaction interfaces
• Audit, privacy, and managerial settings

Current Development Status

AuraCoreCF is actively evolving as a local-first desktop system, with certain features already in production while others remain under active development. The primary functionality is currently optimized for Windows, with additional platform integrations scheduled for future updates.

For ongoing research and publications related to AuraCoreCF and coherence-based cognitive systems, visit the official repository.