Coder CLI

AI-powered CLI tool for code analysis, creation, and fixes. This tool integrates with your AI backend to provide intelligent coding assistance right from your terminal.

__      ___  _____  ____  ____  ____ 
\ \    / __)(  _  )(  _ \( ___)(  _ \
 ) )  ( (__  )(_)(  )(_) ))__)  )   /
/_/    \___)(_____)(____/(____)(_)\_)

)

Installation

Install the CLI globally from npm to use the coder-cli command anywhere on your system.

npm install -g @jekyll-studio/coder-cli

Note: This makes two commands available: coder-cli (the primary command) and coder (a shorter alias). The coder alias might not be available if another program on your system is already using it. This documentation will use coder-cli in all examples for consistency.

Configuration

Before using the CLI for the first time, you need to configure it with your AI backend API key.

coder-cli init

This will launch an interactive setup wizard that will ask for your API key and save it in a local configuration file (~/.coder-cli-config.json).

Usage

Here are the main commands available:

init

Initializes or re-configures the CLI tool.

coder-cli init

chat

Starts an interactive chat session. You can provide a local project or a remote Git repository as context for the AI.

# Start a chat session in the current directory
coder-cli chat

# Use a specific local project as context
coder-cli chat --project /path/to/your/project

# Use a remote public Git repository as context
coder-cli chat --repo https://github.com/user/repo.git

# Start a chat session without streaming responses
coder-cli chat --no-stream

redesign

Redesigns a web page from a given URL into new HTML, CSS, and JavaScript files.

# Redesign a web page from a URL
coder-cli redesign https://www.example.com

explain

Gets an AI explanation for a piece of code from a specified file, optionally focusing on a specific line.

# Explain the entire content of a file
coder-cli explain src/utils/helper.ts

# Explain a specific line in a file
coder-cli explain src/components/Button.tsx --line 25

analyze

Performs a high-level analysis of a project, identifying its structure, main technologies, and potential areas for improvement.

# Analyze the project in the current directory with default text output
coder-cli analyze

# Analyze a specific project path and output in JSON format
coder-cli analyze /path/to/your/project -o json

fix

Starts a chat session in "fix" mode to help you resolve a specific issue.

# Fix an issue in the current project directory
coder-cli fix --issue "The login button is not working on the main page."

# Fix an issue in a specific project path
coder-cli fix /path/to/your/project --issue "There is a null pointer exception in the user service."

create

Starts a chat session in "create" mode to generate new code based on a specification.

# Create new code in the current project directory
coder-cli create --spec "Create a REST API endpoint for user registration."

# Create new code in a specific project path
coder-cli create /path/to/your/project --spec "Add a React component for a contact form."

create-script

Generates a new script file based on a given specification and project context.

# Generate an analytics.js script in the current directory
coder-cli create-script analytics.js -s "a JavaScript file to track page views and user interactions"

# Generate a utility script in a specific project path
coder-cli create-script /path/to/your/project/utils/data-fetcher.js -s "a script to fetch data from an external API"

new

Creates a new project with a specified technology stack and features.

# Create a new project with default vanilla web stack
coder-cli new my-project

# Create a Next.js project
coder-cli new my-next-app -t nextjs

# Create a Vite project with specific features
coder-cli new my-vite-app -t vite -s "with TypeScript and Tailwind CSS"

# Create a React project
coder-cli new my-react-app -t react

# Create a project with custom specifications
coder-cli new my-portfolio -t vanilla -s "with HTML, CSS, and JavaScript for a personal portfolio with dark mode"

The new command intelligently generates complete project structures with all necessary files, configurations, and dependencies based on the chosen technology stack. It supports multiple stacks including Next.js, Vite, React, and vanilla HTML/CSS/JavaScript projects.

AI Models Configuration

The backend run different AI models for different purposes, optimized for performance and cost.

Chat Model

• Model ID: google/gemma-3-12b-it
• Purpose: General conversation, answering questions, and non-code related queries. Ideal for quick, interactive responses.
• Usage: Primarily used for /api/chat when the mode is 'chat' or unspecified.
• Characteristics: Optimized for natural conversation and general knowledge queries. Generally the most cost-effective for basic interactions.

Code Model

• Model ID: qwen/qwen2.5-coder-32b-instruct
• Purpose: Code generation, code fixing, and script generation.
• Usage: Used for /api/create, /api/fix, and /api/script endpoints. Also used in /api/chat when the mode is 'create' or 'fix'.
• Characteristics: Optimized for code understanding, generation, and technical problem solving. Offers a good balance of capability and cost for coding tasks.

Advanced Model

• Model ID: deepseek-ai/deepseek-r1-distill-qwen-32b
• Purpose: Complex code analysis, detailed explanations, project generation, and creative web page redesign.
• Usage: Used for /api/analyze, /api/explain, /api/project, and /api/redesign endpoints. Also used in /api/chat when the mode is 'analyze'.
• Characteristics: A larger, more capable model suitable for tasks requiring deeper understanding, creativity, and comprehensive output. This model is generally more expensive per token.