Code Interpreter

A Rust implementation of an AI-powered code interpreter that executes code in multiple languages (Python, Shell, AppleScript) based on natural language instructions using OpenAI-compatible APIs.

Features

• 🤖 Natural language to code execution via OpenAI-compatible LLMs
• 🐍 Multiple language support: Python, Shell, AppleScript
• 🔄 Interactive REPL interface with readline support
• 📝 Syntax highlighting for code blocks
• 🌐 RAG-powered procedure recommendations via Open Procedures API
• 📊 Structured logging to ./logs/app.log
• ⚙️ Flexible configuration via environment variables

Quick Start

Demo

Prerequisites

• Rust 1.70+ (2021 edition)
• Python 3.x (for Python interpreter)
• OpenAI API key or compatible API endpoint

Installation

# Clone the repository
git clone <repository-url>
cd Code-Interpreter

# Build the project
cargo build --release

# Run the application
cargo run

Configuration

The application supports the following environment variables:

Required

• OPENAI_API_KEY or API_KEY: Your OpenAI API key or compatible API key

Optional

• OPENAI_BASE_URL or BASE_URL: Custom API base URL (default: OpenAI's API)

  • Example: https://api.openai.com/v1
  • Useful for OpenAI-compatible services (e.g., Azure OpenAI, local LLM servers)

• OPENAI_MODEL or MODEL: Model name to use

  • Examples: gpt-4, gpt-3.5-turbo, claude-3-opus-20240229

Example Configuration

# Basic setup with OpenAI
export OPENAI_API_KEY='sk-...'
cargo run

# Using a custom endpoint and model
export OPENAI_API_KEY='your-api-key'
export OPENAI_BASE_URL='https://api.your-provider.com/v1'
export OPENAI_MODEL='gpt-4'
cargo run

# Or use shorter variable names
export API_KEY='your-api-key'
export BASE_URL='https://api.your-provider.com/v1'
export MODEL='gpt-4'
cargo run

Usage

Once running, the application provides an interactive REPL where you can enter natural language commands:

> Calculate the factorial of 10 using Python
> List all files in the current directory
> Create a new folder called 'test' and add a README.md file

The AI will generate and execute appropriate code based on your instructions.

Development

Build Commands

# Build for development
cargo build

# Build optimized release version
cargo build --release

# Run tests
cargo test

# Run a specific test
cargo test test_shell_ls

# Check code without building
cargo check