Legacy Code Explainer & Modernization Assistant

A hackathon prototype: paste in legacy COBOL, JCL, or aging Java, and get back (per logical chunk of code):

• a plain-English business logic explanation
• dead-code / risk flags (unreachable branches, stale flags, discontinued products)
• a first-draft modernized translation (Java or Python)
• an honest confidence score
• a list of open questions a human reviewer must verify before trusting the translation

This is a copilot, not an autopilot. The tool is explicitly designed to surface uncertainty rather than hide it — that's the whole point of the "open questions" and confidence score fields.

---

Project structure

legacy-code-assistant/
├── backend/              FastAPI + Claude API
│   ├── main.py            API endpoints
│   ├── chunker.py         splits legacy source into paragraphs/methods/steps
│   ├── claude_service.py  prompt design + Claude API calls
│   └── requirements.txt
├── frontend/             React (Vite)
│   └── src/
│       ├── App.jsx
│       ├── App.css        design system (legacy phosphor -> modern editor)
│       └── components/
└── sample_data/          synthetic COBOL + Java files for the demo

---

Running it locally

1. Backend

cd backend
cp .env.example .env
# edit .env and set ANTHROPIC_API_KEY=sk-ant-...

python -m venv venv
source venv/bin/activate        # Windows: venv\Scripts\activate
pip install -r requirements.txt

uvicorn main:app --reload --port 8000

The API will be live at http://localhost:8000. Check http://localhost:8000/api/health.

2. Frontend

cd frontend
npm install
npm run dev

Open http://localhost:5173. The Vite dev server proxies /api requests to the backend on port 8000 (see vite.config.js), so no CORS setup is needed beyond what's already in main.py.

3. Demo flow

1. Click Load sample → Premium Calc (or Proration Utils).
2. Click Analyze legacy code.
3. Expand a chunk to see the explanation, any risk flags, the legacy/modern split view, the confidence badge, and open questions.

---

Design notes

The UI's split-panel layout is intentional, not decorative: the left panel uses an amber-phosphor terminal palette (legacy), the right panel uses a clean dark-editor palette (modern), and a gradient "migration rail" literally connects the two — visually staging the transformation the tool performs.

Extending this for a real hackathon submission

Ideas if you have extra time:

• Jira integration: auto-file a modernization ticket per chunk via Atlassian Rovo, pre-filled with the explanation + open questions.
• Repo-wide scan: accept a zipped repo, chunk every file, and produce a prioritized "riskiest modules first" report.
• "Explain like I'm new here" toggle: adjust explanation depth for a new hire vs. a senior engineer.