git-smart-squash 2.5.0

Automatically reorganize messy git commit histories into clean, semantic commits

Git Smart Squash

An AI-powered tool that intelligently reorganizes your git commits into logical, reviewable chunks perfect for pull requests.

What It Does

Instead of manually squashing and organizing commits, Git Smart Squash:

1. Analyzes your entire diff between your feature branch and the main branch
2. Uses AI to organize changes into logical, reviewable commits
3. Automatically restructures your git history to match the AI's recommendations
4. Creates clean, conventional commits that are easy for reviewers to understand

Before vs After

Before (messy commits):

* 3a1b2c3 WIP: auth stuff and models
* 4d5e6f7 more changes
* 8g9h0i1 tests and docs

After (AI-organized commits):

* a1b2c3d feat: implement user authentication system
* e4f5g6h test: add comprehensive auth test coverage  
* i7j8k9l docs: update API documentation for auth endpoints

Installation

Install from PyPI (Recommended)

pip install git-smart-squash

Alternative Installation Methods

Using pipx (isolated environment):

pipx install git-smart-squash

From source:

git clone https://github.com/your-username/git-smart-squash.git
cd git-smart-squash
pip install -e .

Quick Start

1. Set up AI Provider

Local AI (Default - Recommended):

# Install and start Ollama
ollama serve
ollama pull devstral

Or use cloud AI:

# For OpenAI
export OPENAI_API_KEY="your-api-key"

# For Anthropic  
export ANTHROPIC_API_KEY="your-api-key"

# For Google Gemini
export GEMINI_API_KEY="your-api-key"

2. Navigate to Your Git Repository

cd /path/to/your/repo
git checkout your-feature-branch

3. Run Git Smart Squash

Dry run (see what it would do):

git-smart-squash --dry-run

Apply the changes:

git-smart-squash

Use short command:

gss --dry-run  # Same as git-smart-squash --dry-run

Usage Examples

Basic Usage

# Preview the reorganization
git-smart-squash --dry-run

# Apply the reorganization  
git-smart-squash

# Use different base branch
git-smart-squash --base develop

# Use specific AI provider
git-smart-squash --ai-provider openai --model gpt-4.1

Advanced Usage

# Use OpenAI (automatically uses gpt-4.1)
git-smart-squash --ai-provider openai

# Use Anthropic (automatically uses claude-sonnet-4-20250514)
git-smart-squash --ai-provider anthropic

# Use Google Gemini (automatically uses gemini-2.5-pro-preview-06-05)
git-smart-squash --ai-provider gemini

# Use specific model
git-smart-squash --ai-provider openai --model gpt-4.1

# Custom configuration
git-smart-squash --config .custom-config.yml

AI Providers

Local AI (Default)

• Provider: Ollama with devstral model
• Cost: Free
• Privacy: Completely local
• Setup: ollama serve && ollama pull devstral

OpenAI

• Provider: OpenAI GPT models
• Cost: Pay per use
• Setup: Set OPENAI_API_KEY environment variable
• Models: gpt-4.1

Anthropic

• Provider: Anthropic Claude models
• Cost: Pay per use
• Setup: Set ANTHROPIC_API_KEY environment variable
• Models: claude-sonnet-4-20250514

Google Gemini

• Provider: Google Gemini models
• Cost: Pay per use
• Setup: Set GEMINI_API_KEY environment variable
• Models: gemini-2.5-pro-preview-06-05

Configuration

Global Configuration

Create ~/.git-smart-squash.yml:

ai:
  provider: local  # or openai, anthropic, gemini
  model: devstral  # or gpt-4.1, claude-sonnet-4-20250514, gemini-2.5-pro-preview-06-05
  
output:
  backup_branch: true

Project Configuration

Create .git-smart-squash.yml in your project root:

ai:
  provider: openai
  model: gpt-4.1

Safety Features

Automatic Backups

• Creates backup branch: your-branch-backup-1703123456
• Preserves original work automatically
• Easy recovery if needed

Recovery Commands

# If something goes wrong, recover your original branch:
git checkout your-branch-backup-1703123456
git checkout your-working-branch  
git reset --hard your-branch-backup-1703123456

Validation

• Checks for clean working directory
• Validates base branch exists
• Confirms changes before applying

How It Works

The 4-Step Process

1. Get Complete Diff: Uses git diff main...HEAD to capture all changes
2. AI Analysis: Sends diff to AI with prompt for logical commit organization
3. Display Plan: Shows proposed commit structure for review
4. Apply Changes: Uses git reset --soft and creates new commits

Technical Details

• Single Python file (~300 lines) for simplicity
• Direct git commands via subprocess
• Rich terminal UI for clear feedback
• Conventional commit message standards
• JSON response parsing with fallback mechanisms

Token Management (Local AI)

For large repositories, the tool automatically:

• Estimates token usage (1 token ≈ 4 characters)
• Sets optimal context size with hard cap at 12,000 tokens
• Adjusts prediction limits up to 12,000 tokens
• Scales timeouts for large requests (up to 600 seconds)

Troubleshooting

Common Issues

Ollama server not running:

ollama serve

Model not available:

ollama pull devstral

Large diffs timeout:

• Break changes into smaller commits first
• Use --base with more recent branch
• Check token limit warnings

API key issues:

# Check environment variables
echo $OPENAI_API_KEY
echo $ANTHROPIC_API_KEY

Error Messages

"No changes found to reorganize":

• Check you're on the right branch
• Verify there are differences from base branch

"Failed to generate commit plan":

• Check AI provider configuration
• Verify network connectivity (for cloud providers)
• Try with smaller diff

Examples

Simple Authentication Feature

Input diff:

+++ b/src/auth.py
+def authenticate(user, password):
+    return user == "admin"

+++ b/tests/test_auth.py  
+def test_authenticate():
+    assert authenticate("admin", "pass")

AI Output:

[
  {
    "message": "feat: add user authentication system",
    "files": ["src/auth.py", "tests/test_auth.py"], 
    "rationale": "Groups authentication implementation with its tests"
  }
]

Complex Multi-File Feature

AI organizes into logical commits:

• feat: implement user authentication core
• feat: add user model and database integration
• test: add comprehensive auth test coverage
• docs: update API documentation for auth endpoints

Contributing

Development Setup

git clone https://github.com/your-username/git-smart-squash.git
cd git-smart-squash
pip install -e .

# Run tests
python -m pytest test_functionality_comprehensive.py

# Run Ollama integration tests
./test_with_ollama.sh

Running Tests

Comprehensive functionality tests:

python test_functionality_comprehensive.py

Real Ollama integration tests:

python test_ollama_integration.py

License

MIT License - see LICENSE file for details.

Support

• Issues: Report bugs and feature requests on GitHub
• Documentation: See FUNCTIONALITY.md for detailed technical specs
• AI Integration: See OLLAMA_INTEGRATION.md for token management details

---

Made with ❤️ for developers who want cleaner git history