/Code-Atlas

A C++ implementation of Open Interpreter, based on llama.cpp. / Open Interpreter 的 C++ 实现，基于 llama.cpp

Primary LanguageC++MIT LicenseMIT

中文

Code-Atlas

Overview

This project is a lightweight C++-based interpreter inspired by Open Interpreter, designed to execute code snippets across multiple languages with support for Markdown rendering. Compared to Open Interpreter, this project requires significantly fewer resources, making it suitable for devices with limited capacity. It leverages llama.cpp for local large model inference and delivers a highly optimized execution environment.

Features

  1. Code Execution:

    • Supports Python, Batch, PowerShell, and Shell/Bash scripts.
    • Highlights executable code blocks in purple and non-executable ones in yellow.

  2. Markdown Rendering:

    • Handles the following Markdown elements:
      • Headings (levels 1–4)
      • Bold, Italic
      • Inline code and block code
    • Outputs are color-coded and formatted for readability.

  3. Local AI Inference:

    • Uses llama.cpp for large language model inference.
    • Compatible with gguf models, including Qwen2.5-Interpreter, which is specifically fine-tuned for this task based on the Qwen2.5-0.5B model.

  4. Resource Efficiency:

    • Consumes significantly less memory and computational resources compared to Open Interpreter.

Usage

  1. Obtain the source code and compile it using a compatible C++ compiler, or download a precompiled binary if available. (Release Page)
g++ -std=c++17 main.cpp constants.cpp codeexecutor.cpp utils.cpp -o code-atlas.exe

  1. Obtain llama.cpp from its releases page. Ensure you have:

    • llama.cli.exe
    • Necessary .dll files
    • A compatible gguf model (e.g., Qwen2.5-Interpreter, fine-tuned for this task)

  2. Place the following files in the same directory:

    • The compiled executable from this project
    • llama.cli.exe
    • Required .dll files
    • Your selected gguf model

  3. Run the program. Note that administrator privileges may be required on some systems.

System Requirements

  • Windows OS
  • llama.cpp executable and dependencies
  • Compatible gguf model (e.g., Qwen2.5-Interpreter, based on Qwen2.5-0.5B)
  • Compiled executable for this project

To-Do List

  • Expand support to platforms beyond Windows (Linux, macOS, etc.).
  • Introduce API support for integration into other systems.
  • Develop a GUI for user-friendly interaction.
  • Allow users to select their preferred programming language for code execution.
  • Expand language support beyond Python, Batch, PowerShell, and Shell/Bash.
  • Add plugin support for extended functionality.
  • Support other AI model APIs, such as OpenAI, for enhanced flexibility.
  • Implement enhanced sandboxing mechanisms for improved security.
  • Extend Markdown rendering capabilities for additional elements.
  • Support voice mode, free your hands.

Inspiration

This project is deeply inspired by Open Interpreter, a truly innovative and creative project that has the potential to change the way we interact with computers. The concept of seamlessly integrating natural language and programming to automate tasks is groundbreaking and has set the foundation for many other ideas. This project seeks to build upon those ideas by implementing them in C++ for improved performance and efficiency on Windows systems, while maintaining the core inspiration of making computer interaction more intuitive and powerful.