You are setting up JarvisQwen — an open-source, self-hosted AI research autopilot — on this computer. Repository: https://github.com/Vector897/JarvisQwen

Work through the steps below. Explain what you're doing in simple, non-technical language as you go. If anything fails, diagnose and fix it yourself; only ask me when a decision is truly mine to make.

1. Check prerequisites: git, Python 3.11+, Node.js 18+. Install anything missing (or give me the download link and wait for me).
2. If a "JarvisQwen" folder already exists here, run `git pull` inside it to UPDATE it to the latest version. Otherwise, clone the repository.
3. Backend: create a Python virtual environment inside ./server and install with: pip install -e ".[dev]" litellm
4. Frontend: run npm install inside ./web
5. Ask me for my Qwen Cloud API key (free to get at https://home.qwencloud.com -> API Keys; it starts with "sk-"). Save it to a file named .env at the project root, containing exactly one line: DASHSCOPE_API_KEY=<my key>. Never echo my full key back to me and never commit it anywhere. If I don't have a key yet, skip this step — the system runs in dry-run mode (zero cost, simulated AI) and I can add a key later in Settings.
6. Start both services: backend with uvicorn app.main:app --port 8000 (from ./server, using the venv), frontend with npm run dev (from ./web). Then open http://localhost:3000 in my browser.
7. Read the initial admin password from ./server/data/admin_password.txt (created on first start) and tell me. Then walk me through, step by step: signing in, filling in "My research focus" in Settings, and adding my first topic subscription.
8. Finally, run the test suite (python -m pytest in ./server) and confirm to me everything is green.

The prompt is in English because AI assistants follow it most reliably that way — but you can talk to your assistant in any language while it works.