AutoCoder is an advanced AI-powered coding assistant designed to transform natural language instructions into operational Python code. This tool integrates voice input capabilities through OpenAI’s Whisper, allowing users to dictate tasks efficiently. With its autonomous execution loop and capability to learn from past mistakes, AutoCoder refines its coding skills over time.

Key Features

Intelligent Code Generation

• Translates plain English descriptions into executable Python code.
• Supports both local models via Ollama and cloud-based models through OpenRouter.
• Utilizes a self-learning mechanism for improved accuracy with each use.

Voice Input (Whisper)

• Enables real-time speech-to-text functionality for seamless coding.
• Supports various model sizes based on user requirements, optimizing for speed or accuracy.

Internet Integration

• Fetches live weather information using Open-Meteo, with no API key required.
• Conducts web searches through the Brave Search API.
• Facilitates GitHub repository cloning and manages conflict-free dependencies automatically.

Learning From Errors

• Maintains a persistent database of mistakes to inform future coding attempts.
• Preserves conversation context and adjusts its approach based on user feedback.

Text-to-Speech Capabilities

• Integrates Windows SAPI5 for natural voice synthesis with adjustable settings.

Creative Mode

• Generates detailed project briefs based on creative tasks prior to coding execution.

Safe Execution Environment

• Ensures a secure code execution process with sandboxing techniques to minimize errors.

Getting Started

Simply enter conversational commands in English, such as:

• "Create a spinning 3D cube with OpenGL"
• "Show me weather data for London in graph form"

Users can select from various models and click "Run AutoCoder" to see code generated in real-time.

For those using voice input, press the spacebar to record your commands and watch AutoCoder efficiently generate code while incorporating learned lessons for enhanced productivity.

For a more creative approach, the "Imagine First" feature allows AutoCoder to expand on ideas before coding, making it an excellent choice for creative projects.

The architecture of AutoCoder includes modules for code execution, internet tools, machine learning operations, and configuration management, ensuring a comprehensive solution for user coding tasks.

Conclusion

AutoCoder stands out as a potent tool for both novice programmers and seasoned developers seeking a blend of natural language processing and coding capabilities. It offers a unique balance of features that support creativity, learning, and efficiency in coding tasks.