AI Coding Assistants (CLI)

Command-line AI coding assistants let you interact with large language models directly in the terminal to write, debug, refactor code, perform Git operations, and more. For developers who prefer terminal workflows, these tools can dramatically boost productivity.

Claude Code

Claude Code is Anthropic’s official CLI coding assistant, powered by Claude models. It can understand entire codebases, execute commands, and edit files, making it one of the most powerful terminal AI programming tools.

Install Claude Code


# Ensure Node.js 18+ is installed
node --version
 
# If Node.js is not installed, install it first
curl -fsSL https://deb.nodesource.com/setup_24.x | sudo -E bash -
sudo apt install -y nodejs
 
# Install Claude Code globally via npm
npm install -g @anthropic-ai/claude-code
 
# Verify installation
claude --version

Basic Usage


# Start Claude Code in your project directory
cd ~/my-project
claude
 
# Direct query mode (non-interactive)
claude "Explain this project's architecture"
 
# Pipe mode -- pass file contents to Claude
cat error.log | claude "Analyze this error log and find the root cause"
 
# Execute a single task and exit
claude --print "Generate a Python .gitignore file"

Interactive Sessions

After starting claude, you enter an interactive session where you can:


> Read all Python files in the src/ directory and summarize the project structure

> Add a logging decorator function to utils.py

> Find all unhandled exceptions and fix them

> Run the tests and fix any failing cases

> Write a Dockerfile to containerize this project

Configuration


# Set API key
export ANTHROPIC_API_KEY="sk-ant-xxxxx"
 
# Write to shell config for persistence
echo 'export ANTHROPIC_API_KEY="sk-ant-xxxxx"' >> ~/.bashrc
source ~/.bashrc

Project-level configuration:

Create a CLAUDE.md file in the project root to define project context and AI behavior rules:


cat > CLAUDE.md << 'EOF'
# Project Description
 
This is a Python Flask web application project.
 
## Tech Stack
- Python 3.12
- Flask 3.x
- PostgreSQL 16
- Redis
 
## Code Standards
- Use Black for code formatting
- Use mypy for type checking
- All functions must have docstrings
- Tests use pytest
 
## Project Structure
- `app/` -- Main application code
- `tests/` -- Test files
- `migrations/` -- Database migrations
EOF

Command Quick Reference


# Enter interactive mode
claude
 
# Resume last session
claude --continue
 
# Use a specific model
claude --model claude-sonnet-4-20250514
 
# Output as plain text (suitable for scripts)
claude --print "Generate a UUID"
 
# View help
claude --help

Tip

Claude Code automatically reads project files to understand code context. In large projects, writing a good CLAUDE.md file helps the AI understand your project structure and coding standards more accurately, significantly improving interaction quality.

Aider

Aider is an open-source terminal AI pair programming tool that supports connecting to multiple models (Claude, GPT-4, local models, etc.) and can directly edit code files and automatically generate Git commits.

Install Aider


# Ensure Python 3.10+ is installed
python3 --version
 
# Method 1: Install via pipx (recommended)
sudo apt install -y pipx
pipx ensurepath
pipx install aider-chat
 
# Method 2: Install via pip inside a virtual environment (PEP 668: no pip into the system Python)
python3 -m venv ~/aider-env
source ~/aider-env/bin/activate
pip install aider-chat
 
# Method 3: Via official install script
curl -LsSf https://aider.chat/install.sh | bash
 
# Verify installation
aider --version

Connect to Models


# Use Claude (set API key)
export ANTHROPIC_API_KEY="sk-ant-xxxxx"
aider --model claude-sonnet-4-20250514
 
# Use OpenAI GPT-4
export OPENAI_API_KEY="sk-xxxxx"
aider --model gpt-4o
 
# Use local Ollama model
aider --model ollama/qwen2.5-coder:14b
 
# Use DeepSeek
export DEEPSEEK_API_KEY="your-key"
aider --model deepseek/deepseek-chat

Basic Usage


# Start Aider in a Git repository
cd ~/my-project
aider
 
# Specify files to edit
aider src/main.py src/utils.py
 
# Use architect mode (design first, then code)
aider --architect

In the Aider interactive interface:


> Add command-line argument parsing to main.py

> /add tests/test_main.py    # Add a file to the session

> /run pytest                 # Run a command

> /diff                       # View what AI changed

> /undo                       # Undo the last AI change

> /help                       # View all commands

Configuration File

Create .aider.conf.yml in the project root:


# .aider.conf.yml
model: claude-sonnet-4-20250514
auto-commits: true
dark-mode: true
stream: true

Note

Aider automatically creates a Git commit for each AI change. If you don’t want automatic commits, start with the --no-auto-commits flag.

GitHub Copilot CLI

GitHub Copilot CLI brings Copilot’s capabilities to the terminal, helping you generate shell commands and explain command meanings.

Installation

The old gh extension install github/gh-copilot extension is deprecated. Install the standalone Copilot CLI via npm instead:


# First ensure you have Node.js (if not installed)
curl -fsSL https://deb.nodesource.com/setup_24.x | sudo -E bash -
sudo apt install -y nodejs
 
# Install GitHub Copilot CLI via npm
npm install -g @github/copilot
 
# Verify installation
copilot --version

Basic Usage


# Start an interactive session
copilot
 
# Pass a prompt directly (the prompt is a positional argument)
copilot "Find all files larger than 100MB in the current directory"

In an interactive session you describe your intent in natural language; Copilot proposes commands or explanations and asks for confirmation before executing.

Tip

GitHub Copilot CLI requires a valid GitHub Copilot subscription, and the first run guides you through signing in to your GitHub account.

Gemini CLI

Gemini CLI is Google’s terminal AI tool, powered by Gemini models, supporting code understanding, file operations, and multimodal capabilities.

Installation


# First ensure you have Node.js (if not installed)
curl -fsSL https://deb.nodesource.com/setup_24.x | sudo -E bash -
sudo apt install -y nodejs
 
# Install Gemini CLI via npm
npm install -g @google/gemini-cli
 
# Verify installation
gemini --version

Configuration and Usage


# Set API key
export GOOGLE_API_KEY="your-api-key"
 
# Or use Google Cloud authentication
# gcloud auth application-default login
 
# Start Gemini CLI
cd ~/my-project
gemini
 
# Direct query
gemini "Explain this project's directory structure"

In interactive mode:


> Analyze code quality of all files under src/

> Help me refactor database.py to use connection pooling

> Check package.json and update outdated dependencies

Gemini CLI Special Features


# Multimodal capabilities -- analyze images
gemini "Describe the UI layout in this screenshot" < screenshot.png
 
# Handle large codebases (Gemini supports very large context windows)
gemini "Summarize all API endpoints in this repository"

OpenCode

OpenCode is an open-source terminal AI coding assistant (the project now lives at github.com/sst/opencode ), supporting Claude, OpenAI, Google, and local models, among others.

Installation


# Method 1 (recommended): official install script
curl -fsSL https://opencode.ai/install | bash
 
# Method 2: install via npm
npm install -g opencode-ai@latest
 
# Verify installation
opencode --version

Configuration

Create a configuration file opencode.json in the project directory or ~/.config/opencode/:


{
  "provider": "anthropic",
  "model": "claude-sonnet-4-20250514",
  "mcpServers": {}
}

Basic Usage


# Set API key
export ANTHROPIC_API_KEY="sk-ant-xxxxx"
 
# Start in your project
cd ~/my-project
opencode

OpenCode provides an IDE-like terminal user interface (TUI) with a file tree, conversation panel, and terminal output.

CLI Tool Comparison

Feature	Claude Code	Aider	Copilot CLI	Gemini CLI	OpenCode
Developer	Anthropic	Open source	GitHub	Google	Open source
Default model	Claude	Multi-model	Copilot models	Gemini	Multi-model
Local models	Limited	Supported	Not supported	Not supported	Supported
File editing	Direct edit	Direct edit	Direct edit	Direct edit	Direct edit
Git integration	Supported	Auto-commit	Supported	Supported	Supported
Command execution	Supported	Supported	Supported	Supported	Supported
Interface	REPL	REPL	REPL	REPL	TUI
Pricing	API billing	API billing	Subscription	API billing	API billing

Environment Preparation Summary

Before using these CLI tools, prepare the following environment:


# Install Node.js (required for Claude Code, Gemini CLI, Copilot CLI)
curl -fsSL https://deb.nodesource.com/setup_24.x | sudo -E bash -
sudo apt install -y nodejs
 
# Install Python and pipx (pipx recommended for Aider)
sudo apt install -y python3 python3-pip python3-venv pipx
pipx ensurepath
 
# Create API key configuration file
cat >> ~/.bashrc << 'EOF'
 
# AI API Keys
export ANTHROPIC_API_KEY=""
export OPENAI_API_KEY=""
export GOOGLE_API_KEY=""
EOF
source ~/.bashrc

Note

All CLI tools send your code to cloud APIs for processing (unless using local models). When working with sensitive or private code, make sure you understand each tool’s data privacy policy. Using local models (e.g., via Aider + Ollama) completely avoids sending data externally.