Deep Research Agent

A full-stack AI research agent app for deep research using multiple LLMs, with a modern chat UI and real-time streaming results.

Features

• Multi-LLM Research: Select and query multiple large language models (ChatGPT, Gemini, Grok, etc.) in parallel.
• Leading Model Summarization: Choose a leading model to synthesize and conclude based on all model outputs.
• Streaming Results: See model responses and summary/conclusion streamed live as they are ready.
• Rich Chat UI: React + Vite frontend with Markdown rendering, copy-to-clipboard, and responsive design.
• Async Python Backend: FastAPI backend with async API calls, error handling, and CORS support.
• Secure API Key Management: API keys are stored in a separate, git-ignored file for security.
• Web Search Integration: Instructs LLMs to use the provided Brave web search MCP server when a web search for current events is needed.

Project Structure

research-agent/
├── backend/              # FastAPI backend
│   ├── main.py           # Main API logic
│   ├── .env              # API keys (excluded from git)
│   ├── requirements.txt  # Python dependencies
│   └── README.md         # Backend usage
├── frontend/             # React + Vite + TypeScript frontend
│   ├── src/              # Frontend source code
│   ├── index.html        # App entry (uses research.png as favicon)
│   ├── research.png      # App icon
│   └── README.md         # Frontend usage
├── Makefile              # Start/stop/restart both backend and frontend
├── .gitignore            # Excludes venv, node_modules, .env, etc.
└── README.md             # (This file)

Quick Start

1. Backend (Python/FastAPI)

cd backend
uv venv
source .venv/bin/activate
uv pip install -r requirements.txt
# Create .env and add your API keys
uvicorn main:app --reload --host 0.0.0.0 --port 8000

2. Frontend (React/Vite)

cd frontend
npm install
npm run dev

3. All-in-one (from project root)

make start

• The backend runs on http://localhost:8000
• The frontend runs on http://localhost:5173

API Keys

• All API keys are stored in backend/.env (excluded from git).
• You must provide your own keys for OpenAI, Gemini, Grok, etc.

Example .env file

OPENAI_API_KEY=your-openai-key-here
GEMINI_API_KEY=your-gemini-key-here
GROK_API_KEY=your-grok-key-here
BRAVE_SEARCH_API_KEY=your-brave-search-key-here

Testing

Backend (FastAPI)

Run from the project root or backend folder:

pytest backend/test/

Frontend (React)

Run from the frontend folder:

cd frontend
npm test

• Backend tests use pytest and httpx's AsyncClient for endpoint and logic testing.
• Frontend tests use @testing-library/react and Jest for UI/component testing.

Security & Best Practices

• Never commit your API keys (backend/.env is git-ignored).
• The Python virtual environment (venv/) and frontend node_modules/ are also git-ignored.
• Use best practices for async API calls, error handling, and user experience (see .github/copilot-instructions.md).

License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.

---

For more details, see the README.md files in the backend/ and frontend/ folders.