Auto Changelog Generator

A Python CLI tool that automatically generates changelogs from git diffs and staged changes using local AI models. Categorizes changes (feat, fix, docs, breaking), groups related changes, and produces human-readable release summaries following the Conventional Commits format.

Features

• Git Diff Analysis: Analyze staged and unstaged changes from git repositories
• AI Categorization: Categorize changes using local LLM models (Ollama/LM Studio)
• Conventional Changelog: Generate markdown changelogs following Conventional Commits specification
• Multiple Output Formats: Support for markdown, JSON, and GitHub/GitLab release notes
• Git Hook Integration: Install hooks for automatic changelog generation on commits
• Configuration: YAML-based configuration with environment variable overrides

Installation

From Source

# Clone the repository
git clone https://7000pct.gitea.bloupla.net/7000pctAUTO/auto-changelog-generator.git
cd auto-changelog-generator

# Install dependencies
pip install -e .

# Or install with dev dependencies
pip install -e ".[dev]"

Using pip

pip install changeloggen

Dependencies

• Python 3.10+
• GitPython >= 3.1.40
• Typer >= 0.9.0
• Rich >= 13.7.0
• Pydantic >= 2.5.0
• Requests >= 2.31.0
• PyYAML >= 6.0.1

Quick Start

Basic Usage

Generate a changelog from staged changes:

changeloggen generate

Generate with a specific model:

changeloggen generate --model llama3.2

Include unstaged changes:

changeloggen generate --all

Output Formats

Markdown changelog (default):

changeloggen generate --output markdown

JSON output for programmatic consumption:

changeloggen generate --output json

GitHub/GitLab release notes:

changeloggen generate --output release

Generate Release Notes

Create release notes for a version:

changeloggen release --version 1.0.0

Configuration

Configuration File

Create a .changeloggen.yaml file in your project root:

ollama_url: http://localhost:11434
model: llama3.2
temperature: 0.3
output_format: markdown
include_unstaged: false

Or in ~/.config/changeloggen/config.yaml for user-wide settings.

Environment Variables

Variable	Description	Default
CHANGELOGGEN_OLLAMA_URL	LLM API URL	http://localhost:11434
CHANGELOGGEN_MODEL	Model name	llama3.2
CHANGELOGGEN_CONFIG	Config file path	-
CHANGELOGGEN_NO_COLOR	Disable colors	-

Commands

generate

Generate changelog from git diffs.

changeloggen generate [OPTIONS]

Options:

• --output, -o: Output format (markdown, json, release, commit-message)
• --model, -m: LLM model to use
• --all, -a: Include unstaged changes
• --version, -v: Version string for changelog
• --output-file, -f: Write output to file

release

Generate release notes for GitHub/GitLab.

changeloggen release [OPTIONS]

Options:

• --version, -v: Version for release notes
• --output-file, -f: Write output to file
• --model, -m: LLM model to use

install-hook

Install git hooks for automatic changelog generation.

changeloggen install-hook <hook-type> [OPTIONS]

Options:

• --model, -m: LLM model to use
• --branches, -b: Comma-separated branches to run on

Hook types:

• prepare-commit-msg: Run before commit message editor
• commit-msg: Run after commit message is written

remove-hook

Remove installed git hook.

changeloggen remove-hook <hook-type>

list-hooks

List all installed changeloggen hooks.

changeloggen list-hooks

config

Manage configuration.

# Show current configuration
changeloggen config show

# Update configuration
changeloggen config set --model custom-model --temperature 0.5

check

Check system requirements and connectivity.

changeloggen check

version

Show version information.

changeloggen version

Git Hook Integration

Install Hooks

Install prepare-commit-msg hook for all branches:

changeloggen install-hook prepare-commit-msg --model llama3.2

Install for specific branches only:

changeloggen install-hook prepare-commit-msg --branches main,develop

Remove Hooks

changeloggen remove-hook prepare-commit-msg

Output Formats

Markdown Changelog

# Changelog v1.0.0 (2024-01-15)

## ✨ Features
- **feat(auth):** add user login functionality

## 🐛 Bug Fixes
- **fix(api):** fix memory leak in connection pool

## 📝 Documentation
- **docs(readme):** update installation instructions

---

### Summary
This release adds user authentication, fixes a memory leak, and updates documentation.

JSON Output

{
  "version": "1.0.0",
  "generated_at": "2024-01-15T10:30:00Z",
  "summary": "This release adds user authentication.",
  "changes": [
    {
      "type": "feat",
      "scope": "auth",
      "description": "add user login functionality",
      "file_path": "src/auth/login.py",
      "breaking_change": false,
      "confidence": 0.95
    }
  ],
  "breaking_changes": [],
  "contributors": ["@developer1"]
}

Release Notes

## Release v1.0.0

### ✨ Features
- **(auth)** add user login functionality

### 🐛 Bug Fixes
- **(api)** fix memory leak in connection pool

### Contributors
Thank you to our contributors:
- @developer1
- @developer2

---
**This release adds user authentication and fixes a memory leak.**

Development

Running Tests

# Run all tests
pytest tests/ -v

# Run with coverage
pytest tests/ --cov=src/changeloggen --cov-report=term-missing

Linting

# Check code style
ruff check src/changeloggen/

# Auto-fix issues
ruff check src/changeloggen/ --fix

Type Checking

mypy src/changeloggen/ --strict

Requirements

• Git: Must be installed and available in PATH
• Local LLM: Ollama or LM Studio running with REST API
• Python: 3.10 or higher

LLM Server Setup

Ollama

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Start Ollama
ollama serve

# Pull a model
ollama pull llama3.2

LM Studio

1. Download LM Studio from https://lmstudio.ai
2. Start the local server
3. Load a model

Error Handling

Error	Solution
Git repo not found	Initialize git: git init
No staged changes	Stage files: git add <files>
LLM not running	Start Ollama/LM Studio
Invalid config file	Validate YAML syntax
API connection failed	Check LLM server URL

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.