CodeArtisan is a cutting-edge Command Line Interface (CLI) tool designed to leverage Large Language Models (LLMs) for enhancing code quality across various programming languages. It offers intelligent, context-aware auto-fixing of linting issues and refactoring suggestions, ensuring your codebase adheres to the latest best practices and standards.

Install CodeArtisan globally using npm to start improving your codebase:

npm install -g @bxav/code-artisan

Run the init command to set up CodeArtisan in your project. This creates a .codeartisan directory in your project root, containing a config.yml for configuration and an examples directory for custom setups:

code-artisan init

Your project structure will then include:

ProjectRoot/
└── .codeartisan/
    ├── config.yml
    └── examples/

The config.yml file within the .codeartisan directory has been enhanced to suit your project's needs. You can specify the model type, name, and options such as temperature for the model's responses. This allows for a more tailored and precise code refactoring and linting experience.

# OpenAI API
model:
  type: OpenAI
  name: gpt-4-1106-preview # Example: gpt-4-1106-preview, ...
  options:
    temperature: 0.5

# Uncomment to use Mistral API
# model:
#   type: Mistral
#   name: mistral-small

# Uncomment to use a Local LLM
# model:
#   type: Ollama
#   name: codellama # Example: codellama:13b, ...

experts:
  react:
    pattern: .tsx, .jsx
    role: Senior React Developer
    codingStyles:
      - path: ./.codeartisan/react-coding-styles.md
    examples:
      - path: ./.codeartisan/examples/reactExample1.tsx
      - path: ./.codeartisan/examples/reactExample2.tsx
  nestjs:
    pattern: .service.ts, .module.ts, .controller.ts
    role: Senior Nodejs Developer, expert in NestJS
    codingStyles:
      - path: ./.codeartisan/nestjs-coding-styles.md
      - path: ./.codeartisan/nodejs-coding-styles.md
    examples:
      - path: ./.codeartisan/examples/example1.controller.ts
      - path: ./.codeartisan/examples/example1.service.ts
      - path: ./.codeartisan/examples/example1.module.ts

smart-corrector is a command within CodeArtisan aimed at automatically resolving linting issues in your code. It uses the expert roles, coding styles, and examples you've defined to tailor its fixes to your project's specifications.

Automatically fix linting issues throughout your project with:

code-artisan smart-corrector

This scans all relevant files in the current directory, defaulting to Git to identify changed files if no commit diff is specified.

Specify files or folders for linting fixes:

• For individual files:

  code-artisan smart-corrector path/to/your/file.tsx another/path/to/file.tsx

• For folders:

  code-artisan smart-corrector path/to/your/folder/

The smart-corrector supports several options to customize its operation, including --commit-diff for Git commit diffs, --expert for expert-specific suggestions, and --config to specify an alternative configuration file:

code-artisan smart-corrector --commit-diff HEAD~1..HEAD
code-artisan smart-corrector --expert react
code-artisan smart-corrector --config path/to/your/config.yml

Combine multiple options for a comprehensive linting process, including the --config parameter to use a custom configuration file:

code-artisan smart-corrector path/to/your/file.tsx --expert react --config path/to/your/config.yml

To fully enable CodeArtisan's functionality with Large Language Models, you need to set specific environment variables:

• For using OpenAI models, set the OPENAI_API_KEY environment variable with your OpenAI API key.
• For using Ollama models, set the OLLAMA_BASE_URL environment variable to specify the base URL of your Ollama service. This is crucial for the CLI to communicate with your local or hosted Ollama instance.

Example of setting environment variables:

export OPENAI_API_KEY='your_openai_api_key_here'
export OLLAMA_BASE_URL='http://localhost:11434' # Or the URL of your hosted Ollama service

Ensure these variables are correctly set in your environment to leverage the respective models' capabilities within CodeArtisan for code refactoring and linting.