Skip to content

Latest commit

 

History

History
133 lines (108 loc) · 5.82 KB

File metadata and controls

133 lines (108 loc) · 5.82 KB

Product Requirements Document (PRD)

Agentic Linux Assistant (Native OS AI Integration)

Overview

The Agentic Linux Assistant is a system-native AI assistant baked directly into a custom Ubuntu distribution. This is not an application but a core OS feature that provides users with intelligent system interaction through a modern, overlay-style interface. The assistant leverages a local LLM (GPT-OSS-20B via Ollama) to enable natural language control of the entire system, creating a truly agentic computing experience.


Goals

  • Create a truly agentic operating system where AI assistance is a native feature, not an add-on.
  • Provide seamless natural-language control over all system functions and applications.
  • Deliver a modern, intuitive overlay interface accessible instantly via global hotkey (Ctrl+Space).
  • Enable zero-setup experience - works immediately upon OS boot without configuration.
  • Build foundation for advanced autonomous system management and task execution.

Target Users

  • Developers and power users seeking AI-augmented computing workflows.
  • Organizations wanting intelligent, autonomous desktop environments.
  • Early adopters interested in next-generation human-computer interaction.
  • Users with GPU-capable hardware for optimal LLM performance.

Key Features

  1. Modern Overlay Interface

    • Glassmorphism-styled overlay popup with blur effects
    • Centered screen positioning with smooth animations
    • Chat-style conversation interface with streaming responses
    • Adaptive sizing and system-native theming
  2. System-Native Integration

    • Systemd service running at boot across all user sessions
    • Global hotkey: Ctrl+Space works in any application
    • Desktop environment agnostic (GNOME, KDE, XFCE compatible)
    • Zero configuration - works immediately after OS installation
  3. Local LLM Engine

    • Pre-bundled Ollama with GPT-OSS-20B model
    • GPU-accelerated inference with automatic hardware detection
    • Streaming responses for real-time conversation experience
    • Context-aware system interaction
  4. Unrestricted System Control

    • Natural language execution of any system command or task
    • Application launching and window management
    • File system operations and system monitoring
    • Direct subprocess execution based on AI interpretation
  5. ISO Integration

    • Baked into custom Ubuntu ISO as core OS feature
    • Pre-installed models and optimized system services
    • Automatic GPU driver installation and optimization
    • No post-install setup required
  6. Advanced Capabilities

    • Session-persistent conversations
    • System context awareness (current directory, running apps)
    • Multi-user support with per-user contexts
    • Power management integration

Technical Architecture

  • Base OS: Ubuntu 22.04 LTS (Custom ISO)
  • Installation Location: /usr/share/agentic-assistant/ (system-native)
  • Runtime Dependencies:
    • Python 3.11+
    • GTK4 + libadwaita (modern UI framework)
    • pynput (global hotkey detection)
    • asyncio (non-blocking operations)
    • Ollama (pre-bundled)
  • System Integration:
    • systemd service (agentic-assistant.service)
    • Debian package for ISO integration
    • Casper hooks for first-boot optimization
  • Models: GPT-OSS-20B (pre-downloaded in ISO)
  • GPU Support: CUDA/ROCm with automatic detection

User Experience

  1. User presses Ctrl+Space anywhere in the system.
  2. Modern overlay window appears with smooth animation and blur effects.
  3. User engages in natural conversation:
    • "Open Firefox and navigate to GitHub"
    • "Show me system performance and free up some memory"
    • "Create a new Python project in ~/dev with git initialization"
  4. AI streams response in real-time with contextual understanding.
  5. System commands execute automatically based on AI interpretation.
  6. Conversation persists during session for follow-up questions.
  7. Window dismisses with Escape or click outside, maintaining context.

Non-Functional Requirements

  • Performance: Sub-second response times with GPU acceleration; smooth 60fps animations.
  • Reliability: Robust error handling; graceful degradation to CPU inference if needed.
  • Security: System-level access with intelligent command interpretation and execution.
  • Scalability: Efficient memory usage; automatic model optimization based on hardware.
  • Usability: Zero-configuration experience; intuitive modern interface following OS design language.
  • Maintainability: Modular architecture enabling easy model updates and feature additions.

Future Enhancements

  • Multimodal Interaction: Voice input/output, screen understanding, gesture control.
  • Advanced Automation: Complex workflow execution, scheduled task management.
  • Learning Capabilities: User behavior adaptation, personalized command shortcuts.
  • Network Intelligence: Web research, remote system management, cloud integration.
  • Developer Tools: Code generation, debugging assistance, project scaffolding.

Deliverables

  1. Core System Components:

    • agentic-daemon - Main systemd service with hotkey detection
    • overlay-gui - Modern GTK4 popup interface with glassmorphism styling
    • ollama-client - LLM communication and streaming response handler
    • system-integration - Command execution and OS interaction layer
  2. System Integration:

    • Debian package (.deb) for proper dependency management
    • systemd service files for boot-time initialization
    • Desktop environment integration (GNOME, KDE, XFCE)
    • Custom ISO with pre-bundled models and optimized configuration
  3. Documentation & Tooling:

    • Developer documentation for extending capabilities
    • ISO build scripts and customization tools
    • Performance tuning guides for different hardware configurations