Getting started

UniQuick lets you ask an AI app to build a small web app and put it live at a real Illinois link — in one prompt. No servers to set up, no accounts to wire together, no API keys. Here's what your app gets for free, then how to ship your first one in a few minutes.

What your app can do (without you building it)

Every UniQuick app gets five things for free. You don't write the code for these yourself, but knowing what's possible helps you ask for the right thing:

It knows who's visiting. Everyone is already signed in with their Illinois account, so your app knows each visitor's name and netid — you never build a login. (A sign-up sheet already knows who signed up.)
It remembers things, shared by everyone. Your app can save data and show it to every visitor — and it sticks around. (Poll votes, guestbook messages, a leaderboard — saved once, everyone sees it.)
It accepts file uploads. Visitors can upload a photo or PDF and your app gets a link to display or keep it.
It updates live for everyone. Changes show up for all visitors at once with no refresh, and your app can see who's on the page right now. (Live results as people vote, "who's here", shared cursors.)
It has built-in AI. Your app can summarize text, generate questions, or give writing feedback — no API key, no separate AI account. It runs gpt-4o-mini on Azure OpenAI inside the University of Illinois tenant, and Azure OpenAI does not use your prompts or responses to train any model. Each visitor gets their own daily allowance (~100,000 tokens/day), so a whole class can share one app without draining a single budget — and you're never charged for your visitors' AI use.

Because the AI is built in, you never need to put an API key (OpenAI, Anthropic, etc.) into your app — and you shouldn't: an app is just files that anyone at Illinois can read. If an AI tool ever tries to paste a key in, tell it to use UniQuick's built-in AI instead. Your AI app reads the full technical reference at /llms.txt.

1. Choose your AI app

UniQuick works best from an app on your computer that can make files and run the deployment command for you. If you've only used AI in a browser tab, start with one of these desktop apps:

Best first choices:

Codex Desktop — a good first choice if you already use ChatGPT or Codex.
Claude Desktop — a good first choice if you already use Claude.

Editor options if you already work in a code editor:

Cursor — an AI-first editor that can build and deploy for you.
VS Code with GitHub Copilot or the Claude / Codex extensions.
Windsurf — another AI-native editor for working with project files.

Command-line tools are for people already comfortable with a terminal: Codex CLI, Claude Code, or Gemini CLI.

All of these can use UniQuick the same way: read /llms.txt, build a plain static site, and deploy it with your token.

On a phone? The apps you build work great on phones — your visitors sign in, add data, and can even upload a photo straight from the camera. Building an app is easiest on a computer (Mac, Windows, or Linux), especially with a desktop app. Advanced users can also work from a phone through a cloud workspace or Android terminal, but a computer is the smoothest first run.

Some setup paths use Node.js because UniQuick runs through npx. If your AI app says npx is missing, install Node.js once (free) from nodejs.org. UniQuick itself needs no download and no repo access.

2. Create a deploy token

Open quick.disruptionlab.illinois.edu/token in your normal browser and sign in with your @illinois.edu account.
Create a token, label it (e.g. claude-code laptop), and copy it immediately — it is shown once. It looks like qk_…. You'll paste it into your AI app in the next step, so keep it handy.
Treat it like a password; revoke it from the same page any time.

Do not ask your AI app to handle Illinois sign-in. It may use a different browser profile, get stuck on MFA, or fail on a popup. You sign in; then the token lets the app deploy.

3. Paste this prompt into your AI app

This is the beginner path. Replace qk_… with the token you just created:

"Build a single-page UniQuick app for me. Read https://quick.disruptionlab.illinois.edu/llms.txt. Create a plain static site with index.html and any needed assets. Include the UniQuick SDK as the first script in the page head. Use a small local mock if needed so the page can be previewed on localhost. Deploy it with UniQuick using the token below. Use the token only for deployment. Do not put it inside the app code or any public file. Do not share the app to the public gallery unless I ask. Site name: seminar-rsvp My deploy token is below: qk_…"

Site names carry your netid automatically, so asking for seminar-rsvp lands your app at /s/<your-netid>-seminar-rsvp/. Manage or delete your sites at /my, in your normal browser.

4. Advanced setup

If you already know how to set environment variables, you can set the token as UNIQUICK_TOKEN instead of pasting it into the chat.

Setting it by hand depends on your system — export UNIQUICK_TOKEN=qk_… on Mac/Linux, set UNIQUICK_TOKEN=qk_… in Windows Command Prompt, or $env:UNIQUICK_TOKEN="qk_…" in Windows PowerShell. The one-line setup below sets it for you.

Add the UniQuick tools. In Claude Code, one line:

claude mcp add uniquick --env UNIQUICK_TOKEN=qk_… -- npx -y -p uniquick uniquick-mcp

Using Codex? Add the same npx -y -p uniquick uniquick-mcp command as an MCP server in your Codex config (how).

Or skip setup and just ask your AI app in plain English — it will figure out the rest from the reference:

"Use the UniQuick CLI (run it with npx -y uniquick) and read https://quick.disruptionlab.illinois.edu/llms.txt for how it works. My deploy token is in UNIQUICK_TOKEN. Then build me an app."

Just want to look under the hood? The API also works with plain curl (built into Mac, Windows, and Linux) — see /llms.txt. The CLI + MCP are open-source on npm (the uniquick package); the platform repo is private — email vishal@illinois.edu for source access.

Ground rules

Apps are visible to anyone at Illinois with the link — not listed publicly, not private either.
Only you can change or delete your sites; data writes are attributed to real accounts.
Static pages only — no user code runs on the server. Uploads capped at 5MB/file, 40MB/deploy.
Not an approved store for FERPA-protected records or sensitive data.
Runs entirely on University of Illinois Azure infrastructure (App Service, Postgres, Blob storage, and Azure OpenAI), paid for by the Gies Disruption Lab — free for you to build and host on.

Have feedback, an idea, or an app to show off? Post it on the feedback wall (itself a UniQuick app). Questions, repo access, or a site behaving oddly → vishal@illinois.edu.