Neverclick – Keyboard-driven mouse control using local computer vision

Name: Neverclick – Keyboard-driven mouse control using local computer vision
Availability: InStock
Author: lazov

by lazov·Jul 21, 2026·1 point·0 comments

Visit Project View on HN

AI Analysis

●●●BangerWizardrySolve My Problem

Computer vision beats accessibility APIs for speed and works in every app.

Strengths

•CV-based detection works universally where accessibility APIs fail or don't exist.
•Runs on 10-year-old hardware with instant hint rendering — no GPU required.
•Born from real RSI recovery story, not a solution looking for a problem.

Weaknesses

•Windows-only at launch excludes macOS users who also suffer from RSI.
•Computer vision approach may struggle with dynamic or animated UI elements.

Post Description

Hello,

Neverclick is a desktop application that I created when I was suffering from RSI. It uses local computer vision to detect clickable UI elements on your screen so that you can use your keyboard to perform mouse actions.

Most software in this space relies on accessibility APIs but Neverclick uses computer vision. This means that it works in every app. Additionally, it also produces really good bounding boxes around text so it's great for editing text and coding. And counterintuitively, it runs significantly faster than the traditional accessibility api approach.

It's 100% offline. It's not a wrapper around ChatGPT or Claude or any other online AI service. The computer vision system runs on your computer. I've optimized it is as best as I can, the hints should show up instantaneously even on 10 year old hardware. You don't need a powerful gpu to run it.

Also I swear on my life that it wasn't "vibe coded", I've been working on this since long before the recent AI revolution.