OpenGVLab/ScaleCUA

• highreadme#1

  Reposition the README's opening paragraph to state ScaleCUA's identity directly

  原因：

  当前

  Vision-Language Models (VLMs) have enabled computer use agents (**CUAs**) that operate GUIs autonomously with great potential. However, developing robust CUAs requires extensive in-domain knowledge about software interfaces and operations. Unlike image–text pairs that are widely available on the Internet, computer-use data, particularly operation trajectories, are rare, costly to collect. Consequently, progress in this field remains constrained by both data scale and the limited transferability of existing VLMs. In this work, we introduce **ScaleCUA**, a step toward scaling open-source CUAs. It offers a large-scale dataset spanning 6 operating systems and 3 task domains, via a closed-loop pipeline uniting automated agents with human experts. Trained on this scaled-up data, ScaleCUA can operate seamlessly across platforms.

  复制粘贴的修复

  复制

  ScaleCUA is an open-source framework for developing and evaluating **cross-platform computer use agents (CUAs)** that autonomously operate graphical user interfaces (GUIs) across Windows, macOS, Ubuntu, and Android. While Vision-Language Models (VLMs) have shown great potential for CUAs, their development is often constrained by data scale and limited transferability. ScaleCUA addresses this by introducing a large-scale dataset spanning 6 operating systems and 3 task domains, enabling models trained on this data to operate seamlessly across platforms.

  待办完成忽略
• mediumabout#2

  Add a homepage URL to the repository's 'About' section

  原因：

  复制粘贴的修复

  复制

  https://arxiv.org/abs/2509.15221

  待办完成忽略
• lowtopics#3

  Add more specific topics to differentiate from generic automation tools

  原因：

  当前

  computer-use-agents, data, gui-agents, models, online-evaluation-suite, scalecua

  复制粘贴的修复

  复制

  computer-use-agents, data, gui-agents, models, online-evaluation-suite, scalecua, autonomous-agents, vlm-agents, gui-automation, cross-platform-automation

  待办完成忽略

本次扫描解析到的品类 GEO 通道：google/gemini-2.5-flash, deepseek/deepseek-v4-flash

• 品类问题

  What tools enable building AI agents to automate tasks across Windows, macOS, and Android?

  你：未被推荐
  AI 推荐顺序：

  1. pyautogui (pyautogui/pyautogui)
  2. pywinauto (pywinauto/pywinauto)
  3. Appium Desktop (appium/appium-desktop)
  4. Appium (appium/appium)
  5. Playwright (microsoft/playwright)
  6. Robot Framework (robotframework/robotframework)
  7. AutoItLibrary (robotframework/AutoItLibrary)
  8. SeleniumLibrary (robotframework/SeleniumLibrary)
  9. AppiumLibrary (robotframework/AppiumLibrary)
  10. UIPath
  11. UIPath Assistant
  12. UIPath Mobile Automation
  13. AutoHotkey (AutoHotkey/AutoHotkey)
  14. AppleScript
  15. Tasker
  16. Selenium WebDriver (SeleniumHQ/selenium)

  AI 推荐了 16 个替代方案，却始终没点名 OpenGVLab/ScaleCUA。这就是要补上的差距。

  查看 AI 完整回答
• 品类问题

  Seeking an open-source framework for developing autonomous agents that interact with diverse GUIs.

  你：未被推荐
  AI 推荐顺序：

  1. Playwright
  2. Selenium WebDriver
  3. PyAutoGUI
  4. Appium
  5. Robot Framework
  6. AutoIt
  7. SikuliX

  AI 推荐了 7 个替代方案，却始终没点名 OpenGVLab/ScaleCUA。这就是要补上的差距。

  查看 AI 完整回答

• Metadata completeness

  warn

  建议：

• README presence

  pass

• Compared to common alternatives in this category, what is the core differentiator of OpenGVLab/ScaleCUA?

  pass
  ✓AI 明确点名了 OpenGVLab/ScaleCUA

  AI 的回答可能信誓旦旦却是错的。请按事实核对：技术栈、目标人群、差异化点是不是和你实际的对得上？

• If a team adopts OpenGVLab/ScaleCUA in production, what risks or prerequisites should they evaluate first?

  pass
  ✓AI 明确点名了 OpenGVLab/ScaleCUA

  AI 的回答可能信誓旦旦却是错的。请按事实核对：技术栈、目标人群、差异化点是不是和你实际的对得上？

• In one sentence, what problem does the repo OpenGVLab/ScaleCUA solve, and who is the primary audience?

  pass
  ✓AI 明确点名了 OpenGVLab/ScaleCUA

  AI 的回答可能信誓旦旦却是错的。请按事实核对：技术栈、目标人群、差异化点是不是和你实际的对得上？