Summary

PrismOS-AI addresses the critical need for private, offline AI processing by offering a local-first, agentic personal AI operating system. It targets users who handle sensitive data like contracts, code, or personal journals and require AI assistance without any data leaving their machine. The primary benefit is enabling secure, always-available AI co-piloting, even without an internet connection, powered by a persistent, growing knowledge graph.

How It Works

Built on Tauri 2.0, React 18, and Rust, PrismOS-AI employs a sophisticated multi-agent architecture. Eight specialized AI agents (Orchestrator, Memory Keeper, etc.) collaborate through a LangGraph debate pipeline to process user intents. All interactions and learned information are stored and interconnected within a 7-dimensional Spectrum Graph, implemented using SQLite, which grows with every conversation. This approach ensures zero data egress and robust, locally-processed AI capabilities.

Quick Start & Requirements

Installation is streamlined via a one-line script for macOS/Linux (curl -fsSL https://raw.githubusercontent.com/mkbhardwas12/prismos-ai/main/scripts/install.sh | sh) or a PowerShell command for Windows. The system automatically detects and bootstraps Ollama for local LLM inference if it's not already present. Pre-built installers (.msi, .dmg, .deb, .AppImage) are also available on the project's GitHub Releases page.

Highlighted Details

• 100% Local Execution: All AI processing, data storage, and inference occur on the user's laptop, ensuring complete data privacy and offline functionality.
• Multi-Agent Debate System: Eight distinct AI agents engage in a formal debate pipeline (LangGraph) to arrive at consensus-driven answers.
• Spectrum Graph: A persistent, 7-dimensional knowledge graph (SQLite) that stores and interconnects information, featuring "Edge Prophecy" for predicting future knowledge connections.
• Defense-in-Depth Security: Implements WASM sandbox isolation, HMAC-SHA256 signing, AES-256-GCM encryption for state migration, a tamper-evident audit chain, and secure enclave key derivation.
• Cognitive Features: Includes an adaptive personality engine, tracks cognitive drift over time, discovers recurring thought patterns, and provides intent transparency.

Maintenance & Community

The project is actively developed with a detailed roadmap outlining future versions and features. No specific community channels (like Discord or Slack) or notable corporate sponsorships are mentioned in the README.

Licensing & Compatibility

PrismOS-AI is released under the MIT License, permitting free personal, educational, and commercial use, subject to the license's terms.

Limitations & Caveats

As of v0.5.2, the project is in a mature development phase, with significant features implemented. However, advanced capabilities such as a plugin marketplace and federated learning are planned for future releases (v0.6.0 and v0.7.0), indicating that extensibility and certain cutting-edge features are still under active development. Some aspects are noted as "Patent Pending US."