👻 Git Ghost: Your Code Never Truly Dies, An AI That Haunts Your Git History

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine

What I Built

👻 Git Ghost — Your Code Never Truly Dies.

Git Ghost is a horror-themed AI companion that lives inside your Git repository.

Every developer has deleted code, abandoned features, forgotten commits and old implementations. But what if you could bring those memories back?

I wanted to interpret the theme "Ghost in the Machine" literally by creating a ghost that haunts your codebase and remembers its past.

Features:

• 👻 Ghost Awakening – A spooky terminal sequence that introduces the ghost when a repository is connected.
• 🪦 Code Graveyard – Discover deleted files and when they disappeared.
• 🕰️ Haunted Timeline – Explore commit history, authors and changes.
• 🔮 The Séance – Chat with an AI ghost about your repository's history.
• 💀 Repository Scanner – Explore branches, commits and tracked files.

The idea is to make exploring Git history feel less like reading logs and more like communicating with the past of your codebase.

Demo

Git Ghost currently runs locally.

Local URL: http://127.0.0.1:5000

Code

GitHub Repository: https://github.com/Rohitisavailable/git-ghost

Built using Python, Flask, GitPython, JavaScript and Ollama.

How I Built It

I wanted to experiment with open-weight AI models and local inference instead of depending on paid AI APIs.

Technology Stack:

• Python – Core application logic.
• Flask – Backend and API endpoints.
• GitPython – Extracting repository history, commits and deleted-file events.
• Ollama – Running AI models locally.
• Qwen 2.5 3B – Open-weight model powering the ghost's conversations.
• HTML, CSS and JavaScript – Frontend, animations and haunted dashboard.

How it works:

1. The user provides a local Git repository path.
2. GitPython scans the repository and extracts historical information.
3. The backend prepares repository metadata.
4. Relevant information is passed to Qwen through Ollama.
5. The model responds as Git Ghost using the repository information as context.

I also implemented an offline fallback so the application can still demonstrate basic functionality when Ollama isn't running.

Git remains the source of truth. The AI interprets the available history rather than inventing commit hashes, authors or deletion reasons.

Why Does Open Innovation Matter?

Open innovation made it possible to experiment with local AI without depending on paid inference APIs.

Using Qwen 2.5 3B through Ollama allowed me to run the model on my own machine and integrate it directly into my application.

It also gave me more control over how the model behaves and how repository information is provided to it.

I wanted to explore how open-weight models could be used beyond conventional chatbots, particularly in creative developer tools.

The Spirit of It

There's something interesting about Git history.

We spend hours writing code, modifying features, deleting files and rewriting entire modules. Eventually, we forget why certain decisions were made.

But Git remembers.

I liked the idea of turning that history into something atmospheric.

Instead of simply displaying:

File deleted in commit a81f92

Imagine a ghost telling you:

"You thought you could erase it? I remember that file."

That was the experience I wanted to create.

Because in Git, nothing really disappears. It just gets buried. 👻

Built with Python, open-source tools and a little Halloween spirit. 👻

---