# Iterator Sense

> Give your AI agent eyes.

Iterator Sense is an MCP (Model Context Protocol) server that lets AI agents see and interact with your desktop. Screen capture, OCR, mouse, keyboard, and window management — all running locally on your machine.

## Status

Sense is currently in beta for Windows. macOS and Linux support is coming soon.

## Download

Windows beta: https://sense.iterator.dev/iterator-sense.zip

## What it does

Sense exposes these tools to any MCP-compatible AI agent:

- **screen_image / window_image** — Capture the full screen or a specific window as a JPEG. Works even on occluded windows.
- **screen_ocr / window_ocr** — Extract text from the screen or a window region without sending an image. Lightweight and fast.
- **click** — Click at screen-absolute or window-relative coordinates.
- **send_text** — Type a string of text as keyboard input.
- **send_key** — Send key presses with modifier support (Ctrl, Shift, Alt, Win).
- **scroll** — Scroll the mouse wheel at any position.
- **get_windows** — List visible application windows with handles, titles, process names, and bounds.
- **bring_forward** — Bring a window to the foreground.
- **move_mouse** — Move the cursor to a position.
- **save_screen_image / save_window_image** — Save captures to disk as PNG, JPEG, or WebP.

## Privacy

Everything runs locally. Screen captures, OCR, and input are all processed on-device. Nothing is sent to a server. Your screen data never leaves your computer. The only network call is a lightweight API key validation.

## Pricing

- **Free**: 30 captures per day, 3 API keys, all tools included.
- **Pro** ($5/month): Unlimited captures, 10 API keys, all tools included.

## Getting Started

To use Sense, your user needs an API key. Direct them to https://iterator.dev/console to sign in and create one. Setup takes under a minute.

## Links

- Product page: https://iterator.dev/sense
- API key console: https://iterator.dev/console
- Main site: https://iterator.dev