CognOS: AI-Augmented Operating System

CognOS is an AI-augmented operating system that provides natural language command interpretation through local language models. The system translates user intent into real system commands, automates environment setup, and provides an interactive, context-aware interface that makes managing Linux environments more accessible and intuitive.

Project Overview

CognOS offers two deployment models:

• CognOS Prototype: Pre-configured Raspberry Pi OS image for demos and testing
• CognOS Distribution: Complete custom Linux distribution for production deployment

Core Technology

• AI Engine: Mistral 7B (Q4/Q5 quantized) via llama.cpp
• Architecture: Custom shell replacement with AI agent integration
• Tools: Extensible tool system for filesystem, environment, and application management
• Safety: User confirmation and comprehensive logging for all AI actions

---

Repository Structure

CognOS/
├── src/                    # Source code
│   ├── shell/             # cognos-shell executable
│   ├── agent/             # AI agent core
│   ├── tools/             # Agent tools (search_folder, run_command, etc.)
│   ├── ui/                # Desktop overlay (future)
│   └── common/            # Shared utilities (config, logging)
├── config/                # Configuration templates
├── scripts/               # Installation and setup scripts
├── models/                # AI models (downloaded during setup)
├── tests/                 # Test suite
└── docs/                  # Documentation

Core Architecture

Shell Layer

• Custom Shell: cognos-shell executable replaces user's default shell
• Command Interception: Distinguishes natural language from direct shell commands
• AI Integration: Routes natural language to Mistral 7B agent
• Safety Layer: User confirmation for all AI-generated commands

AI Agent

• Local LLM: Mistral 7B Q4/Q5 via llama.cpp
• Tool System: Extensible tool registry with safety controls
• Function Calling: Structured JSON responses with tool execution
• Logging: Comprehensive audit trail of all AI decisions

Tool Ecosystem

• search_folder: Filesystem navigation and discovery
• list_options: Multi-choice selection interface
• run_command: Safe shell command execution
• create_env/switch_env: Python environment management
• Extensible: Plugin architecture for custom tools

User Interface

• Primary: Enhanced shell with natural language processing
• Optional: Desktop overlay with global hotkey activation
• Integration: ChatGPT API for general-purpose queries

---

Quick Start

Development Setup

# Clone repository
git clone https://github.com/bpolania/CognOS.git
cd CognOS

# Setup environment (requires Python 3.8+)
make setup

# Download AI models
make models

# Build and test
make build
make test

Installation

# Install system-wide
make install

# Set as default shell
chsh -s /usr/local/bin/cognos-shell

Development

• Development Setup: Complete setup guide for Mac + Raspberry Pi development
• Project Structure: Detailed codebase organization
• Changelog: Development history and session logs

Configuration

• System: /etc/cognos/config.json
• User: ~/.config/cognos/config.json
• Logs: ~/.local/share/cognos/cognos.log
• Models: ./models/mistral-7b-q4.gguf

Contributing

1. Read the appropriate deployment documentation (Prototype or Distribution)
2. Set up development environment using Development Setup
3. Follow the development phases and contribute to the current phase
4. Update changelog for code changes (documentation-only changes excluded)

---

CognOS: Making Linux environments more accessible through natural language AI interaction