Gitra

Clone any GitHub repo and ask AI questions about the code.

Import a repository, browse files like in an IDE, and get instant explanations from AI. No more context switching between browser and editor.

---

Quick Start

Desktop App (Easiest)

git clone https://github.com/ashfromsky/gitra.git
cd gitra
pip install -r requirements.txt
python desktop.py

The app will open automatically at http://127.0.0.1:46789

Web Version

git clone https://github.com/ashfromsky/gitra.git
cd gitra
pip install -r requirements.txt

# Copy example env file
cp .env.example .env

# Add your Gemini API key to .env
# Get it free at https://aistudio.google.com/app/apikey

# Start server
uvicorn app.gitra:app --reload --port 8000

Visit http://localhost:8000

Docker

git clone https://github.com/ashfromsky/gitra.git
cd gitra
docker-compose up

Visit http://localhost:8000

---

What It Does

Gitra is a code exploration tool. You paste a GitHub URL, and it clones the repo locally. Then you can:

• Browse files in a clean IDE-style interface
• Select any piece of code and ask AI about it
• Get bug reports and improvement suggestions
• View repository statistics (stars, forks, languages)

Think of it as "GitHub meets AI assistant" but running locally.

---

Features

Code Browser

• Syntax highlighting for 20+ languages
• File tree with icons
• Markdown preview
• Image viewer
• Multi-tab editor

AI Assistant

• Analyze entire files or selected code
• Ask specific questions about logic
• Get bug detection and refactoring tips
• Works with Google Gemini API

Repository Stats

• GitHub metadata (stars, forks, watchers)
• Language breakdown
• Commit activity
• Topic tags

UI

• Dark/light themes (Gruvbox palette)
• Vim-inspired status line
• Keyboard shortcuts
• Tab management like VSCode

---

How It Works

1. Import - Paste any public GitHub URL
2. Browse - Files are cloned to downloaded_repos/
3. Analyze - AI reads your code and system prompt from app/routes/ai/genai_gemini/
4. Ask - Select code, click "Ask AI", get explanations

The AI panel uses an iframe with local message passing. No code leaves your machine except API calls to Gemini.

---

Configuration

API Keys

⚠️ Security Note: API keys are stored in plaintext in user_config.json. Do not commit this file. For production use, consider encryption.

You need a free Google Gemini API key. Get one at aistudio.google.com/app/apikey

Option 1: Settings UI

1. Open Gitra
2. Click the gear icon (top right)
3. Paste your API key
4. Click "Save Key"

Option 2: Environment Variable

# Add to .env file
GEMINI_API_KEY=AIzaSyD...your-key-here

Option 3: Config File Keys are stored in user_config.json (auto-created, gitignored)

GitHub Token (Optional)

For repository stats, add a GitHub token:

# Add to .env
GITHUB_TOKEN=ghp_your-token-here

Without this, stats will show limited data due to API rate limits.

---

Project Structure

gitra/
├── app/
│   ├── gitra.py                    # FastAPI app
│   ├── config_manager.py           # Settings handler
│   ├── routes/
│   │   ├── basic/combined/repo.py  # File browsing
│   │   ├── ai/base_response.py     # AI endpoint
│   │   └── stats/repo_stats.py     # GitHub stats
│   └── utils/exporter.py           # Repo to markdown
├── templates/
│   ├── index.html                  # Landing page
│   ├── view_repo.html              # IDE interface
│   └── repo_stats.html             # Stats dashboard
├── static/
│   ├── ai_chat.html                # AI panel iframe
│   └── *.svg                       # Logos
├── desktop.py                      # Desktop app launcher
└── downloaded_repos/               # Cloned repos (gitignored)

---

Keyboard Shortcuts

Key	Action
Cmd/Ctrl + K	Open AI assistant
Alt + W	Toggle word wrap
Esc	Close active tab

---

Limitations

Current version supports:

• Public repositories only (private repos planned)
• Single AI provider (Gemini)
• Basic git operations (clone only, no pull/push)

Not supported yet:

• Multiple repositories open at once
• Search across files
• Git history browsing
• Code editing (read-only for now)

---

Why I Built This

I got tired of switching between GitHub, my editor, and ChatGPT just to understand a new codebase. Wanted something that does all three in one window.

---

Roadmap

v1.1 - Core improvements

• Private repository support (OAuth)
• Search in files
• Multiple AI providers (Claude, local LLMs)
• Git operations (pull, checkout branches)

v1.2 - Advanced features

• Code editing with syntax checking
• Diff viewer
• Multi-repo workspace

---

Development

Running Tests

# Install dev dependencies
pip install pytest pytest-asyncio

# Run tests
pytest tests/ -v

Building Desktop App

# Install PyInstaller
pip install pyinstaller

# Build executable
pyinstaller desktop.py --onefile --windowed

# Output in dist/

---

Contributing

Pull requests are welcome. For major changes, open an issue first.

Areas that need help:

• Private repo authentication
• Alternative AI providers (Ollama, Claude)
• Search implementation
• Mobile UI improvements

---

Tech Stack

• Backend: FastAPI, Uvicorn
• AI: Google Gemini 2.5 Flash
• Frontend: Vanilla JS (no frameworks)
• Styling: CSS variables (Gruvbox theme)
• Desktop: pywebview
• Syntax: highlight.js
• Icons: Inline SVG

---

License

MIT License - use it however you want.

See LICENSE for details.

---

Links

• GitHub: github.com/ashfromsky/gitra
• Issues: github.com/ashfromsky/gitra/issues
• API Keys: aistudio.google.com/app/apikey

---

Built in Kyiv. Made for developers who read more code than they write.