🏛️ Legacy Code Archaeologist

An AI-powered system that lets you talk to the *history* of any GitHub repository — not just its current code.

Legacy Code Archaeologist is a production-grade Retrieval-Augmented Generation (RAG) tool designed to analyze how and why a codebase evolved over time.
Instead of reading static snapshots, it mines actual Git diffs, allowing you to ask high-impact questions like:

• “Who introduced the timeout bug..?”
• “Why was the authentication logic rewritten in 2021?”
• “When did this API contract change?”

Built for scalability, accuracy, and real-world engineering workflows, now supercharged by Groq's lightning-fast inference engine.

---

🚀 Key Features

🔍 Deep Git History Intelligence

• Parses real commit diffs (added/removed lines) — not just file snapshots.
• Understands why code changed, not just what changed.

⚡ High-Performance Mining Engine

• Generator-based architecture enables O(1) memory usage.
• Handles massive repositories (Linux, React, Kubernetes) efficiently.

🧠 AI-Powered Q&A

• Uses Groq API (LLaMA-3.3-70B) for lightning-fast, highly accurate contextual reasoning.
• Powered by ChromaDB for semantic search over commit history.

🕰️ Time Machine Mode

• Switch between:
  • Recent History (Fast Scan)
  • Deep Excavation (Full Repo Analysis)

📊 Audit-Grade Reporting

• Automatically generates PDF audit reports of chat sessions.
• Ideal for compliance, audits, and engineering reviews.

🧹 Smart Resource Management

• Shallow Git clones (--depth) for 99% faster fetches.
• Auto-cleanup of cloned repos and vector databases.

---

🧱 Tech Stack

Backend

• Python 3.12+
• GitPython (Custom Mining Engine)
• ChromaDB (Local Vector Store)

AI Engine

• Groq API (llama-3.3-70b-versatile via OpenAI compatible endpoint)

Frontend

• Streamlit

DevOps / Tooling

• uv (Fast Python package manager)
• python-dotenv

---

🧭 System Architecture

┌──────────────┐
│   Git Repo   │
└──────┬───────┘
       ↓
┌─────────────────────┐
│  Custom Miner Engine│  ← Generator-based diff processing
└──────┬──────────────┘
       ↓
┌─────────────────────┐
│   ChromaDB Vector   │  ← Semantic indexing
└──────┬──────────────┘
       ↓
┌─────────────────────┐
│      Groq API       │  ← Lightning-fast LLM reasoning
└──────┬──────────────┘
       ↓
┌─────────────────────┐
│   Streamlit UI      │  ← Chat + Time Controls
└─────────────────────┘

---

🖥️ Demo / Usage

1. Paste any public GitHub repository URL
2. Select:
   • Fast Mode → Recent commits only
   • Deep Mode → Full historical analysis
3. Provide your Free Groq API Key via the UI (if not set in .env).
4. Ask natural language questions:
   • “Why was this function refactored?”
   • “Who changed the authentication logic?”
5. Export a PDF audit report if needed.

---

⚙️ Installation

1️⃣ Clone the repository

git clone https://github.com/kartik0905/git-archaeologist.git
cd git-archaeologist

2️⃣ Install dependencies

pip install -r requirements.txt

(Or use uv pip install -r requirements.txt for faster installation)

3️⃣ Configure environment variables

Create a .env file:

GROQ_API_KEY=your_groq_api_key_here

4️⃣ Run the application

streamlit run app.py

---

🗂️ Project Structure

.
├── app.py              # Streamlit UI & user interaction
├── miner.py            # Core mining engine (diff parsing, batching)
├── requirements.txt
└── pyproject.toml

---

🧠 Why This Project Stands Out

• Designed like a real production system, not a demo.
• Handles large-scale repositories efficiently.
• Solves a real developer pain point — understanding legacy code.
• Uses Groq to eliminate LLM latency, making log parsing instantaneous.

---

📜 License

MIT License

---

⭐ If You Like This Project

Star the repository and feel free to contribute or fork it for your own tooling.

---

Built with engineering discipline, not just prompts.