Git Craft

Sann Sarr — Thu, 15 Oct 2026 00:45:50 +0000

This is a submission for the "Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle" (https://dev.to/challenges/hacktoberfest-week2-2026-10-12)

What I Built

Git Craft: A Little AI Assistant for Better Git Commits 🛠️

Writing code is one thing. Writing a good commit message for it is another.

Sometimes, after making several changes, I find myself spending more time figuring out how to describe them in a commit message than I expected. That small frustration inspired Git Craft, a Python command-line tool that uses AI to draft concise Git commit messages from your changes.

Instead of switching to a browser-based AI chatbot, you can run Git Craft directly in your terminal.

Here's what it does:

Examines staged changes, or unstaged tracked changes when nothing is staged.
Uses an Ollama model to generate a concise commit message.
Includes recent commit subjects to provide additional context.
Prints the suggested message so you can review it before committing.

My goal is to make the everyday Git workflow a little smoother, especially for developers who prefer working in the terminal.

Who could pay for it?

Git Craft is currently an open-source project, not a paid product. Its potential audience includes individual developers, students, and small development teams who want a more convenient commit-message workflow.

A future paid offering could focus on team-specific commit conventions, shared configurations, or integrations. These are possible future directions, not existing features.

Demo

GitHub repository: "ssannssarr/git-craft" (https://github.com/ssannssarr/git-craft)

Git Craft is a CLI tool, so a short terminal recording showing a staged change and its generated commit message would make a great demo.

Code

Explore the source code and contribute:

"GitHub: ssannssarr/git-craft" (https://github.com/ssannssarr/git-craft)

Built with Python and distributed under the MIT License.

How I Built It

I built Git Craft as a Python CLI utility around Ollama's local model-serving interface.

The basic workflow is straightforward:

Git Craft inspects the current repository's changes.
It gathers the relevant diff and recent commit subjects for context.
It sends that information to a model through an OpenAI-compatible API endpoint.
The model drafts a commit message, which Git Craft prints for me to review.

The default configuration uses a local Ollama server and the "llama3" model identifier. The base URL and model can also be configured, allowing the tool to adapt to different compatible endpoints.

I wanted to explore how open-source AI could become a practical part of an everyday developer workflow, rather than just another chatbot interface.

Why Does Open Innovation Matter?

For a tool like Git Craft, open innovation matters because developers should have choices about how their AI tools work.

Using Ollama makes it possible to run a compatible model locally instead of depending entirely on a hosted AI API. This gives developers more control over their model choice, deployment, and where inference happens.

Open-source development also lets other developers inspect the implementation, adapt the workflow, suggest improvements, and contribute features that fit their own needs.

Git Craft is a small project, but it reflects something I value about open innovation:

AI becomes more useful when developers can experiment with it, understand it, and build on top of it.

My Agent Session

I haven't included an agent-session recording in this submission. If I capture a session using DevRelay, I'll add it here.

Prize Categories

Git Craft is a Python CLI tool that uses Ollama for AI-powered commit-message generation. I'll select only the official prize categories for which the project meets the stated requirements.

I don't want to claim eligibility for partner-specific awards without implementing the required integrations or confirming that the project meets their criteria.

DEV Community: Sann Sarr

Git Craft