parcky-cli 2026.6.0a4

AI-powered git commit and PR creation tool

AI CLI 🤖

AI-powered Git commits and Pull Requests, done right.

AI CLI is a command-line tool that uses Google Gemini to generate high-quality commit messages and Pull Requests based on your actual code changes — automatically, consistently, and following best practices.

Spend less time writing commits and PRs. Spend more time coding.

---

✨ What AI CLI does

• 🤖 Generate Conventional Commit messages from real diffs
• 📋 Create Pull Requests automatically, including:
  • What changed
  • Why it changed
  • How to test
• 🚀 Optional auto-push after commit
• 📁 Smart commit-all, grouping changes by context
• 🔗 Branch-based PR creation
• 🎯 Strict Conventional Commits compliance
• ⚙️ Fully configurable prompts per project or globally
• 🧪 Type-safe, tested, production-ready

---

🚀 Quick Start (2 minutes)

Requirements

• Python 3.13+
• Git
• Google Gemini API key
• GitHub CLI (gh) (optional, for PR creation)

Install

git clone <repository-url>
cd ai-cli
uv sync
uv tool install .

Configure

Get your API key at: https://makersuite.google.com/app/apikey

Then configure your environment (example):

export GEMINI_API_KEY="your_api_key_here"

Common usage

# AI-powered commit (with push)
ai-cli smart-commit

# Commit without pushing
ai-cli smart-commit --no-push

# Commit and create a PR
ai-cli smart-commit --pr

# Commit all changes grouped by context
ai-cli smart-commit-all

# Create PR from current branch
ai-cli create-pr

# Create PR against another base
ai-cli create-pr --base develop

---

🧠 How Pull Requests are generated

AI CLI analyzes:

• all commits in the branch
• all changed files (including additions/removals)
• diff stats and curated patches
• configuration and public interface changes

It then generates a PR with a stable structure:

• What was done
• Why it was done
• How to test
• Risk & impact (when applicable)

This avoids shallow or misleading PR descriptions.

---

🎨 Customizing AI Prompts

All AI behavior is customizable via prompts.json. Lookup order:

1. ./prompts.json (project-level)
2. ~/.config/ai-cli/prompts.json (global)
3. Built-in defaults

Example:

{
  "commit_message": {
    "description": "Commit message prompt",
    "prompt": "Generate a Conventional Commit message..."
  },
  "pull_request": {
    "description": "PR prompt",
    "prompt": "Generate a PR title and body..."
  }
}

You can:

• Change language
• Enforce stricter formats
• Customize PR structure
• Adapt prompts per repository

---

⚙️ Configuration

Environment variables:

• GEMINI_API_KEY — Gemini API key (required)
• GEMINI_MODEL — Model to use (default: gemini-2.5-flash)
• MAX_DIFF_SIZE — Max diff size sent to AI (default: 10000)
• DEFAULT_BRANCH — Default git base branch (default: main)
• DEBUG — Debug mode (default: false)

---

🏗️ Architecture (for contributors)

AI CLI follows a clean, layered architecture:

src/ai_cli/
├── core/            # Domain models, interfaces, exceptions
├── services/        # Business logic
├── infrastructure/  # Git, AI, GitHub integrations
├── cli/             # Command-line interface
└── config/          # Configuration management

The design emphasizes:

• separation of concerns
• testability
• deterministic AI inputs
• safe fallbacks when AI is unavailable

---

🔧 Development

# Install dev dependencies
uv sync --group dev

# Format & lint
uv run ruff format src tests
uv run ruff check src tests

# Type checking
mypy src

# Run tests
pytest

---

🔄 Workflow

1. Analyze git changes
2. Generate commit or PR with AI
3. Show preview
4. User approves
5. Commit / Push / PR

AI is always assistive, never destructive.

---

🤝 Contributing

Contributions are welcome!

• Fork the repo
• Create a feature branch
• Make changes
• Run tests
• Commit with Conventional Commits
• Open a PR

---

📄 License

MIT License — see LICENSE.

Made with ❤️ by Parcky.