lazycommit-cli 0.2.0

Automatic git commit and push tool with LLM-generated commit messages

LazyCommit

I got tired of writing commit messages. So I vibe coded this to let AI do it for me.

If you're also too lazy to type git add, git commit -m "...", and git push every single time, this tool is for you.

What it does

Runs one command. AI looks at your changes, writes a commit message, commits everything, and pushes it. Done.

New features:

• 🎨 Beautiful terminal output with rich formatting, tables, and progress indicators
• 🤔 Interactive mode - review and edit AI-generated messages before committing
• 📊 Repository stats - view commit history and repository information
• ⏪ Easy undo - safely undo the last commit
• 🛡️ Smart git state detection - handles merge conflicts, detached HEAD, and other edge cases
• ⚙️ Config management - easily view and edit configuration

Installation

pip install lazycommit-cli

Or install from GitHub:

pip install git+https://github.com/nrmlthms/lazycommit.git

Setup

Get an OpenAI API key and set it:

export OPENAI_API_KEY="your-key-here"

Usage

LazyCommit now uses subcommands for better organization and functionality.

Commit Changes (Main Command)

lazycommit commit                    # Auto-detect, commit, and push
lazycommit commit --no-push          # Commit without pushing
lazycommit commit -m "my message"    # Use your own message
lazycommit commit --dry-run          # See what would happen
lazycommit commit -v                 # Verbose output
lazycommit commit --safe-mode        # Create backup branch

Interactive Mode: By default, LazyCommit will show you the AI-generated message and ask for confirmation:

• Press y to accept
• Press n to reject
• Press e to edit in your default editor

To skip the interactive prompt, provide a message with -m or set interactive_mode: false in config.

Manage Configuration

lazycommit config --show              # Show current config
lazycommit config --edit              # Edit config file
lazycommit config --get model         # Get specific value
lazycommit config --set model --value gpt-4  # Set value
lazycommit config --reset             # Reset to defaults

View Repository Stats

lazycommit stats              # Show stats for last 10 commits
lazycommit stats -n 20        # Show stats for last 20 commits

Displays a beautiful table with:

• Commit hash
• Author
• Date
• Message
• Repository summary (branch, total commits)

Undo Last Commit

lazycommit undo              # Undo last commit (keep changes staged)
lazycommit undo --hard       # Undo and discard all changes
lazycommit undo -f           # Skip confirmation prompt

Safe by default: Uses soft reset to keep your changes staged. Use --hard only if you're sure you want to discard changes.

Configuration

LazyCommit supports configuration via a .lazycommitrc file in your home directory.

Creating a config file

Create ~/.lazycommitrc with your preferences:

{
  "model": "gpt-4o-mini",
  "temperature": 0.7,
  "max_tokens": 100,
  "max_message_length": 500,
  "max_context_files": 10,
  "max_diff_lines": 20,
  "push_by_default": true,
  "safe_mode_by_default": false,
  "verbose_by_default": false,
  "interactive_mode": true,
  "show_progress": true,
  "cache_enabled": true,
  "offline_mode": false
}

Configuration options

LLM Settings:

• model: OpenAI model to use (default: "gpt-4o-mini")
• base_url: Custom API endpoint for OpenRouter or other providers (default: null)
• temperature: LLM temperature for message generation (default: 0.7)
• max_tokens: Maximum tokens for commit message (default: 100)

Commit Message Settings:

• max_message_length: Maximum commit message length in characters (default: 500)
• max_context_files: Maximum files to include in LLM context (default: 10)
• max_diff_lines: Maximum diff lines per file to include (default: 20)
• max_input_tokens: Maximum estimated input tokens before warning (default: 8000)

Behavior Settings:

• push_by_default: Whether to push by default (default: true)
• safe_mode_by_default: Enable safe mode by default (default: false)
• verbose_by_default: Enable verbose output by default (default: false)
• interactive_mode: Prompt to review/edit generated messages (default: true)
• show_progress: Show spinner/progress indicators during operations (default: true)

Retry Settings:

• api_retry_enabled: Enable automatic retry on API errors (default: true)
• api_max_retries: Maximum number of retry attempts (default: 3)
• api_initial_retry_delay: Initial delay between retries in seconds (default: 1.0)

Cache Settings:

• cache_enabled: Enable commit message caching (default: true)
• cache_max_age_days: Maximum age of cached messages in days (default: 30)
• cache_max_entries: Maximum number of cached messages (default: 100)

Offline Mode:

• offline_mode: If true, only use cache, never call API (default: false)

Configuration priority

Settings are applied in this order (highest priority first):

1. Command-line arguments (e.g., --model gpt-4)
2. Environment variables (e.g., OPENAI_API_KEY, LAZYCOMMIT_MODEL)
3. Config file (~/.lazycommitrc)
4. Default values

Environment variables

• OPENAI_API_KEY: Your OpenAI API key (required)
• BASE_URL: Custom API endpoint (e.g., for OpenRouter)
• LAZYCOMMIT_MODEL: Override the model setting
• LAZYCOMMIT_TEMPERATURE: Override temperature
• LAZYCOMMIT_MAX_TOKENS: Override max tokens
• LAZYCOMMIT_PUSH_BY_DEFAULT: Set to "true" or "false"
• LAZYCOMMIT_SAFE_MODE: Set to "true" or "false"
• LAZYCOMMIT_VERBOSE: Set to "true" or "false"

Using OpenRouter

To use OpenRouter or other OpenAI-compatible APIs, set the BASE_URL environment variable or add base_url to your config file:

Via environment variables:

export OPENAI_API_KEY="your-openrouter-api-key"
export BASE_URL="https://openrouter.ai/api/v1"
export LAZYCOMMIT_MODEL="anthropic/claude-3.5-sonnet"

Via config file (~/.lazycommitrc):

{
  "model": "anthropic/claude-3.5-sonnet",
  "base_url": "https://openrouter.ai/api/v1",
  "temperature": 0.7,
  "max_tokens": 100
}

Then set your OpenRouter API key:

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

Features

🎨 Beautiful Terminal Output

LazyCommit uses rich for beautiful terminal formatting:

• Colored output
• Progress spinners during slow operations
• Formatted tables for stats and config
• Syntax highlighting

🤔 Interactive Message Review

Review AI-generated commit messages before committing:

• See the generated message in a nice panel
• Accept, reject, or edit the message
• Opens your preferred editor (VISUAL/EDITOR env vars)
• Can be disabled with interactive_mode: false in config

🛡️ Smart Git State Detection

LazyCommit detects and handles various git repository states:

• Merge conflicts: Blocks commit and shows conflicted files
• Detached HEAD: Warns but allows commit with suggestion to create branch
• Rebase in progress: Blocks commit until rebase is complete
• Cherry-pick/revert in progress: Blocks commit with helpful suggestions
• Normal state: Commits proceed as expected

📊 Repository Statistics

View your commit history in a beautifully formatted table:

lazycommit stats

Shows:

• Recent commits (hash, author, date, message)
• Current branch
• Total commit count
• Repository path

⏪ Easy Undo

Made a mistake? Undo is simple:

lazycommit undo        # Soft reset (keeps changes)
lazycommit undo --hard # Hard reset (discards changes)

With confirmation prompts and clear warnings for safety.

💾 Smart Caching

LazyCommit caches commit messages to:

• Reduce API calls for identical changesets
• Speed up repeated operations
• Save on API costs

Cache automatically expires after 30 days and is limited to 100 entries.

🔄 Automatic Retry

API calls automatically retry on transient errors:

• Timeout errors
• Connection errors
• Rate limit errors
• 5xx server errors

With exponential backoff and configurable retry limits.

🔒 Safe Mode

Use --safe-mode to:

• Create a backup branch before committing
• Automatically rollback on push failures
• Keep your work safe during risky operations

Why?

Because typing commit messages is easy but boring. This tool saves me from doing minimal work that I'm too lazy to do manually.

Development

# Clone the repo
git clone https://github.com/nrmlthms/lazycommit.git
cd lazycommit

# Install dependencies
pip install -e ".[dev]"

# Run tests
pytest

# Run type checking
mypy lazycommit

License

MIT - Do whatever you want with it