Neverclick uses computer vision so it works in all applications.

Hit Tab to toggle between Window mode and Full Screen mode.

Use Shift + W/A/S/D to switch between windows in Window mode.

• Use Shift + Q and Shift + W to cycle through overlapping windows.
• Use Shift + Z to switch into the window behind the currently active one.

Use the arrow keys (or Shift + J I L K) to snap the target to the edge of the bounding box.

Press a number row key (1, 2, 3, 4, 5, 6, 7, 8, 9) to position the target.

The target is in the center by default but you can move it anywhere you like.

Press ' to enter Dual Select mode to select 2 hints.

Press Shift + ' to enter Multi Select mode to select multiple hints. Press Enter to finish.

Press Ctrl + ' to enter Path mode. Select 2 or more hints. All the hints between them will be selected automatically. Press Enter to finish.

Activate the Text Highlighting Tool, then pick two hints to highlight the text between them.

Useful for dragging files between folders or rearranging the order of tabs in your web browser.

• The bounding box is the rectangular outline of the detected UI element.
• The hint is the label.
• The target is where the mouse action is executed.

The target is placed in the center of the bounding box by default but you can move it.

However, you can customize this if you have a different keyboard layout.

Set up your hotkeys in a GUI.

Configure your keybinds visually on an interactive virtual keyboard.

Neverclick Vision is a computer vision system which finds UI elements based on raw pixel data.

• It's extremely fast (instantaneous even on older hardware)
• It's able to detect text quite well so it's excellent for editing text and coding
• It supports a full screen mode
• Detection settings can be tuned

Neverclick Vision is the default vision mode.

Clairvoyance uses operating system accessibility APIs to find UI elements.

• It's significantly slower than Neverclick Vision
• It can be unpredictable
• It only works on the foreground window (no full screen mode is supported)
• It doesn't detect text in text editors
• Detection settings cannot be tuned
• However, it may produce cleaner bounding boxes for buttons in certain apps

Clairvoyance makes for a good backup to Neverclick Vision.

Press [ when the hints are up to switch into Clairvoyance mode.

Grid mode spaces hints evenly across the screen. It's useful for clicking blank areas, like right-clicking the desktop.

Press ; when the hints are up to switch into grid mode.

• Scrolling
• Window switching via hints
• Moving windows between virtual desktops
• Much more...

Yes. It's free to download and use. There are no accounts, subscriptions, or trials.

For now, yes. Linux and macOS aren't supported yet.

Yes.

Yes, and yes.

Yes. By default the hint labels use your keyboard's language but you can change it in the settings. The settings UI itself is English-only for now.

No. Neverclick installs to Program Files because Windows requires that for it to register as an accessibility tool. This is what lets it perform mouse actions on apps running as administrator.

It's instantaneous. There should be no perceptible delay upon activating it, even on 10 year old hardware.

It's really good, but it does hallucinate occasionally. I'll continue to improve it.

No.

About 40mb.

It depends on your setup: how many monitors you have, their resolution, and your display scaling. On a typical 1080p monitor it's around 200mb. More monitors or higher resolutions will use more.

No. Neverclick doesn't run in a continuous loop. It just sits there and only runs when you activate it. Its CPU usage should be at 0% at all times, except for when you actually use it.

No. Neverclick's computer vision system runs entirely on your own machine. It's not built on ChatGPT, Claude, or any online AI service.

No. Everything runs locally on your machine. Nothing you do ever leaves your computer.

No. It requires an internet connection to install, but once installed, it runs without one. It's 100% offline.

Right now the repo is for issues only. The commits on the repo are a mirror of the real codebase and will be updated occasionally. I uploaded some dummy files so that GitHub recognizes this project as being written in C++.

Yes. It's my daily driver, so I have every incentive to keep improving it and fixing bugs as I hit them.

It's about 95% human made (created by me) and 5% AI. I've worked on Neverclick for many years, predating the AI era. I only started using AI in March 2026. The percentage written by AI will increase over time, but I'm careful not to let it spaghettify the codebase. Additionally, the artwork was predominantly made in Blender (by me) and I created the logo in a vector graphics editor.

Yes. I made Neverclick when I was suffering from RSI.

No. I recovered years ago and I owe it to this software.

It's meant to be an abstract mouse cursor but also the letter "N" when you connect the dot.