// Quick start

Prompy — Quick Start

Prompy turns rough drafts into clear, well-structured prompts — right where you type. It runs on your own AI provider key, so nothing goes through us.

There are two ways to use it: a browser extension (suggests as you type on ChatGPT, Claude, Gemini, and more) and a desktop app (works with any app via the clipboard). You only need one — pick whichever fits how you work.

Step 1 · Get an API key

Prompy uses your own key from one provider. Pick one and grab a key:

Provider Get a key
Claude (Anthropic) console.anthropic.com
GPT (OpenAI) platform.openai.com/api-keys
Gemini (Google) — has a free tier aistudio.google.com/app/apikey
Mistral console.mistral.ai

Your key is stored only on your device and sent straight to that provider. See the Privacy Policy.

Step 2 · Set up the browser extension

  1. Download/clone Prompy.
  2. Go to chrome://extensions, turn on Developer mode (top-right).
  3. Click Load unpacked and select the extension/ folder. (Once published, you'll install it from the Chrome Web Store instead.)
  4. Click the Prompy icon in your toolbar to open the popup.
  5. Choose your AI provider, paste your API key, and click Test connection. A successful test saves it automatically.

Step 3 · Improve your prompts

While typing (the main way)

More ways

Step 4 · Use the desktop app

The desktop app works with any application (including the ChatGPT/Claude desktop apps) through your clipboard.

  1. Launch it: npm run app (or build it — see the README).
  2. In Settings, pick your provider, paste your key, and Test connection.
  3. Playground — paste a draft, click Improve, then Copy the result.

Improve from anywhere with the hotkey

Add conversation context (desktop)

Sandboxed desktop apps can't be read directly, so the clipboard is the bridge.

Use a local model (Ollama / LM Studio) — free & fully private

Prefer no API key and nothing leaving your machine? Point Prompy at a local model server. It works with anything that exposes the OpenAI-compatible API.

Ollama

  1. Install from ollama.com, then pull a model: ollama pull llama3.2
  2. Ollama serves automatically at http://localhost:11434.
  3. In Prompy, choose provider Local, set the Server URL to http://localhost:11434/v1, click Test connection — it lists your models.
  4. Browser extension only: Ollama blocks unknown origins, so allow the extension. Restart Ollama with: OLLAMA_ORIGINS=chrome-extension://* ollama serve (the desktop app needs no such setting).

LM Studio

  1. In LM Studio, download a model and start the Local Server (Developer tab).
  2. Set Prompy's Server URL to http://localhost:1234/v1 and Test connection. LM Studio enables CORS by default, so the extension works out of the box.

Once connected, everything (typing suggestions, Playground, hotkeys) runs against your local model — no key, no cost, no data leaving your device.

Troubleshooting

Questions or issues? Email arhaankhan7b@gmail.com or prompy922@gmail.com.