Larn

Larn is a command-line tool designed to help software engineers improve their coding skills through AI-driven feedback. It analyzes code changes, identifies areas of improvement, and generates personalized programming exercises based on those weaknesses.

Features

• Code Review Automation: Analyzes code changes and provides feedback on quality, bugs, performance, readability, and security.
• Personalized Learning: Identifies areas of struggle based on the feedback and creates programming challenges to help users improve.
• AI Integration: Uses GPT-4o to generate detailed feedback and assignments.

Requirements

To run this project, you will need:

• Rust (version 1.56 or higher)
• Redis (for storing feedback)
• Git (for tracking code changes)
• A Git repository initialized in your project directory.
• An OpenAI API key (for code analysis and challenge generation)

Installation

1. Install Larn via Cargo:

   Run the following command to install Larn globally:

   cargo install larn
   

2. Set up Redis:

   Make sure Redis is installed and running on your local machine.

3. Initialize a Git repository:

   Larn requires an initialized Git repository to track code changes. In the root directory of your project, run:

   git init
   

4. Set up GPT-4 API:

   You need to have access to the GPT-4 API. Set up your environment variable for the API key:

   export OPENAI_API_KEY="your-gpt4-api-key"
   

Usage

Larn provides two main commands: add and learn.

add Command

This command collects code changes from the current directory and retrieves feedback.

larn add

What happens:

• The tool scans the directory for code changes (supported languages: .rs, .py, .js, .cpp, etc.).
• It runs git diff to capture modifications (hence, Git initialization is required).
• The code changes are sent to GPT-4o for review.
• Feedback is stored in Redis.

learn Command

This command generates programming challenges based on previous feedback.

larn learn

What happens:

• It retrieves past feedback from Redis.
• It identifies the areas where you're struggling based on feedback.
• GPT-4 generates programming exercises to improve those areas.
• A text file with assignments is created in the root directory.
• Older feedback is deleted from redis everytime yhis is run.

Example Workflow

1. Make code changes in your project.
2. Run the add command to get feedback on those changes.
3. Once it's time to end the day run the learn command to receive personalized exercises to improve your skills.

Example Commands

# Get feedback on code changes
larn add

# Generate a learning exercise based on feedback
larn learn

Configuration

Larn is configured to work with multiple programming languages, including:

• Rust
• Python
• JavaScript
• C++
• Java
• TypeScript
• Go
• Ruby
• PHP

You can modify the supported file extensions in the source code if needed.

Contributing

Contributions are welcome! If you find any bugs or have feature requests, please open an issue or submit a pull request.

License

Larn is licensed under the MIT License. See the LICENSE file for more details.

Happy coding with Larn!