SeaLin-Assistant

A C-based Linux terminal assistant powered by local AI models that can execute shell commands and search the web to help users with their Linux system tasks.

Overview

SeaLin-Assistant is an intelligent command-line tool that integrates AI capabilities directly into your Linux terminal. It uses the llama.cpp library to run local language models and can:

• Execute shell commands on your Linux system
• Search the web for information
• Provide intelligent responses based on command execution results
• Maintain conversation context throughout the session

Features

• Local AI Processing: Uses GGUF format models (default: Qwen2.5-Coder-3B-Instruct)
• Shell Command Execution: Can run Linux commands and see their output
• Web Search Integration: Search capability for additional information
• Interactive Chat Interface: Conversational interface with the AI assistant
• GPU Acceleration: Supports GPU layers for faster inference

Prerequisites

• Linux operating system
• GCC compiler with C99 support
• Make build system
• CUDA-compatible GPU (optional, for GPU acceleration)

Dependencies

The project uses the following libraries (included in vendor/):

• llama.cpp: For running GGUF format language models
• ggml: Machine learning tensor library
• Standard C libraries: pthread, math

Installation

1. Download a GGUF model under ./models

2. Clone the Repository

git clone https://github.com/alifthi/SeaLin-Assistant
cd SeaLin-Assistant

3. Build the Project

make

4. Install System-wide (Optional)

make install

This installs the binary to /usr/local/bin/SeaLin-Assistant

Usage

Running the Assistant

make run

# Or run directly
./build/SeaLin-Assistant

# If installed system-wide
SeaLin-Assistant

Project Structure

linux-assistant-c/
├── src/                   # Source files
│   ├── main.c             # Main program entry point
│   ├── inference.c        # AI model inference logic
│   ├── states.c           # State management
│   └── utils.c            # Utility functions
├── include/               # Header files
│   ├── config.h           # Configuration constants
│   ├── inference.h        # Inference function declarations
│   ├── states.h           # State structure definitions
│   └── utils.h            # Utility function declarations
├── models/                # AI model files (GGUF format)
├── vendor/                # Third-party libraries
│   ├── include/           # Library headers
│   └── lib/               # Compiled libraries
├── build/                 # Build output directory
└── Makefile              # Build configuration

Model Requirements

• Format: GGUF (GGML Universal Format)
• Recommended: Qwen2.5-Coder series for code generation
• Size: 3B parameters or larger recommended
• Quantization: Q8_0 or higher for better quality

Performance Tuning

• Adjust N_GPU_LAYERS based on your GPU memory
• Modify MAX_MESSAGE_LENGTH for longer conversations
• Use smaller quantized models for faster inference

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test thoroughly
5. Submit a pull request

Acknowledgments

• Built with llama.cpp
• Uses GGML for tensor operations
• Powered by Qwen2.5-Coder language model