資料月份 2026-09

Ui.Vision

uiOpen-source browser and desktop automation running locally, not in the cloud. Vision, OCR, built-in AI, MCP for AI agents and LLMs.

在 Chrome 商店安裝 ↗官網 ↗

據 AlphaMoat 統計,截至 2026-09,Chrome 外掛 Ui.Vision 使用者數為 200,000,環比 +0%,評分 3.9(239 人評價),在全部 9,324 款 AI 瀏覽器外掛中排名第 57 位,在「程式設計開發」分類(601 款)中排名第 3 位。

使用者數 · 2026-09

20萬+0%

評分

★ 3.9239 評價

全站排名(9,324 個)

#57

「程式設計開發」排名

#3/ 601

Ui.Vision 是什麼

Chrome 商店介紹

Open-source browser and desktop automation running locally, not in the cloud. Vision, OCR, built-in AI, MCP for AI agents and LLMs. Open-source browser automation and RPA for Chrome: record macros, scrape data, fill forms, and automate with OCR, image recognition and AI.

Ui.Vision is a free macro recorder and web automation tool that combines classic browser automation with modern computer vision, OCR, and a built-in AI assistant. Record a macro once and replay it as often as you like - locally, deterministically, no coding required.

**NEW in version 10:**

AI Assistant: Describe what you want in plain English ("Create a macro that fills out this form") and the built-in AI builds, runs, and fixes the macro for you - it reads the page, writes the automation, tests it, and repairs what fails. Unlike an AI agent that re-thinks every run, what you get back is a normal macro: yours to keep, edit, and replay a thousand times without AI costs. A free AI tier is included - no API key or account needed. Or bring your own key (Anthropic, OpenRouter) or a local LLM.

JavaScript macros: Macros can now be plain modern JavaScript with a simple uiv.* API - real loops, real if/else, real error handling, breakpoints and step-through debugging. Perfect for CSV-driven work, scraping and conditional logic. Your existing command-table macros keep working exactly as before; nothing is converted.

Claude Code, Cursor, Codex support: Ui.Vision now speaks MCP, so Claude Code and other AI coding tools can build, run and debug macros in your browser while you watch. Everything stays local.

展開其餘商店介紹▾

Browser Vision: New commands click and type like a real user - with visual targeting by image or text (OCR) - directly inside Chrome and Edge and the browser window can stay in the background while the macro runs.

Faster: Script macros run 50-300x faster in loops, and the side panel starts in a fraction of the time.

(1) Browser Automation & Web Macro Recorder

Ui.Vision's visual macro recorder and computer vision commands make automating tasks inside the web browser easy: form filling, web scraping, data extraction, UI testing, and screenshots. Record, edit and run macros right in the browser side panel, next to the page you are automating - the full IDE window is one click away. Existing Selenium IDE scripts can be imported. Conversion help for iMacros scripts is provided, too.

(2) Visual Automation with Image Recognition and OCR

Automate what you can SEE, not just what is in the HTML: click buttons by screenshot, find text on the screen with OCR, and verify page layouts visually. Image recognition and OCR now run directly inside the extension - no additional installs required.

(3) Desktop Automation for Windows, Mac, and Linux

Beyond web browser automation, Ui.Vision can interpret images and text on the desktop, executing actions like clicking, moving, dragging and dropping the mouse, and simulating keyboard inputs.

This desktop automation requires installing the free Ui.Vision XModules, available for Windows, Mac, and Linux. These modules provide Ui.Vision with the necessary capabilities for desktop interaction.

(4) AI Commands & Anthropic Claude Integration

The AI commands allow you to automate complex tasks with a single line of code that would traditionally require hundreds of lines of classic commands. For example, you can teach Ui.Vision to play TicTacToe with just one short "Play this game..." prompt. And the new AI chat assistant in the sidebar creates and repairs complete macros from a plain-language description.

(5) Command Line API

Ui.Vision can be controlled from the command line. Thus it can be used from, and combined with, any programming or scripting language, such as Windows batch files, Linux/Mac shell scripts, Python or PowerShell.

(6) Open-Source

Ui.Vision is Open-Source. The source code is available on Github.

(7) 100% Local Software - Your Data Stays With You

The software does not send any data back to us or any other place. Everything, including image recognition and OCR processing, is done locally on your machine.

The only exception to the "all data is processed locally" rule is if you use the optional AI features. All cloud features are disabled by default - and even for AI you can stay fully local by connecting Ui.Vision to a local LLM. The default computer vision (image recognition and OCR) always runs 100% locally on the machine.

Whether you call it RPA, macro recorder, task automation, web scraping, screen scraping, form filler, or UI testing - if it is repetitive, Ui.Vision can automate it.

**Happy Automating!**

For questions and suggestions, please visit the Ui.Vision community forum at https://forum.ui.vision.