Keeping changelogs up to date is one of those development chores that everyone agrees is important… and everyone forgets to do. Manual changelog curation often falls behind or gets skipped entirely. But what if your Git workflow could automatically generate changelog entries, summarize diffs intelligently, and update your CHANGELOG.md for you?

In this article, we’ll build an intelligent Git hook using Python and OpenAI that auto-generates changelog entries based on Git diffs. You’ll learn how to integrate this into your release process or commit workflow, ensuring that every feature, fix, or improvement is documented clearly and consistently.

The Problem with Traditional Changelog Generation

Most tools rely on conventional commits (feat:, fix:, etc.) or manual editing. These approaches work, but they depend heavily on developer discipline. Common problems:

• Missing entries
• Redundant or unhelpful messages
• Vague descriptions (“updated something”)
• Last-minute copy-paste summaries from PRs

LLMs change the game by being able to:

• Analyze diffs semantically
• Understand file types and changes
• Summarize with consistency and clarity
• Suggest proper categorization

What We'll Build

A pre-commit or pre-push Git hook that:

1. Parses staged or recent diffs
2. Sends them to OpenAI with a prompt to summarize changelog-worthy changes
3. Appends entries to CHANGELOG.md under the correct section

Project Structure

We’ll add a new command to our existing ai-git-hooks CLI tool:

ai_git_hooks/
├── changelog.py      # New!
├── cli.py            # Add `changelog` command

Step 1: The Changelog Generator

Create changelog.py:

import openai
import os
from git import Repo
from datetime import date

openai.api_key = os.getenv("OPENAI_API_KEY")

def get_recent_commits(repo_path=".", count=5):
    repo = Repo(repo_path)
    commits = list(repo.iter_commits('HEAD', max_count=count))
    return commits

def summarize_commits(commits):
    summaries = []
    for commit in commits:
        diff = commit.diff(commit.parents[0] if commit.parents else None, create_patch=True)
        patch_text = "\n".join(
            d.diff.decode('utf-8', errors='ignore') for d in diff if d.diff
        )
        prompt = f"Generate a concise changelog entry for the following diff:\n\n{patch_text}"
        response = openai.ChatCompletion.create(
            model="gpt-4",
            messages=[{"role": "user", "content": prompt}]
        )
        summaries.append(f"- {response['choices'][0]['message']['content'].strip()}")
    return summaries

def update_changelog(summaries, changelog_path="CHANGELOG.md"):
    today = date.today().isoformat()
    header = f"\n## {today}\n"
    body = "\n".join(summaries)

    with open(changelog_path, "a") as f:
        f.write(header)
        f.write(body)
        f.write("\n")

Step 2: Add to CLI

Update cli.py:

from ai_git_hooks.changelog import get_recent_commits, summarize_commits, update_changelog

@cli.command()
@click.option('--count', default=3, help='Number of recent commits to include')
def changelog(count):
    """Auto-generate changelog entries from recent commits"""
    commits = get_recent_commits(count=count)
    summaries = summarize_commits(commits)
    update_changelog(summaries)
    click.echo("✅ CHANGELOG.md updated.")

Run it like this:

ai-git-hooks changelog --count 5

Step 3: Hook It Into Your Workflow

Option A: Post-merge Hook

Create .git/hooks/post-merge:

#!/bin/sh
ai-git-hooks changelog --count 5
git add CHANGELOG.md
git commit -m "chore: update changelog"

Option B: Manual Trigger in Release Script

In your release process:

ai-git-hooks changelog --count 10
git add CHANGELOG.md
git commit -m "docs: update changelog for v1.2.0"

Bonus: Use Conventional Sections

Want to split into sections like Features, Fixes, Refactors? Adjust the prompt:

prompt = f"""
Classify and summarize the following diff into one of these categories:
- Features
- Bug Fixes
- Documentation
- Performance
- Refactoring

Format the response as:

### Features
- Added XYZ

{patch_text}
"""

You can also pre-fill your CHANGELOG.md template with:

## YYYY-MM-DD

### Features

### Fixes

### Refactors

Then have the script insert summaries under the right headers.

Limitations

• LLM summaries may lack technical precision
• Large diffs can hit token limits — chunk intelligently
• Not ideal for massive rebases or one-line commits with many side effects

Conclusion

Changelogs are too important to be forgotten. By integrating OpenAI into your Git tooling, you can automate changelog generation with context, clarity, and consistency — no more stale TODO: fill in changelog notes.

Want to keep leveling up your Git workflow? Next, we’ll explore a post-merge Git hook that explains what just happened — ideal for onboarding or cross-team collaboration.

Find more tools like this at Slaptijack — where developer productivity meets AI innovation.