trycua/cua is a toolkit for building and testing AI agents that operate computers by seeing screens, clicking controls, and typing across operating systems and virtual machines. It serves agent developers, benchmark authors, and teams generating training data for computer-use models. Catalogue add-ons provide ways to operate Cua drivers and related agent workflows.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/trycua/cua/gui-automationnpx skills add trycua/cua --skill gui-automationgit clone --depth 1 https://github.com/trycua/cuaWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/trycua/cua/gui-automation)<a href="https://agentmods.dev/skills/trycua/cua/gui-automation"><img src="https://agentmods.dev/badge/skills/trycua/cua/gui-automation.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00068 | $0.01423 |
| Opus 5 | $0.00034 | $0.00711 |
| Sonnet 5 | $0.00014 | $0.00285 |
| Haiku 4.5 | $0.00007 | $0.00142 |
Grade A, and why
gui-automation scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 162 lines — stays where its author put it; the contents beside it link to each section on GitHub.
GUI Automation
CUA gives you eyes and hands on a real computer: see the screen, move the mouse, click, type, drag, and manage windows — like a human at the keyboard.
Use this skill for visual interaction that can't be done via shell or API.
Setup
cua --version # check install; if missing: pip install cua
# Connect to target (pick one)
cua do switch cloud my-vm
cua do switch docker my-container
cua do-host-consent && cua do switch host # local machine (one-time consent)
ANTHROPIC_API_KEYis optional. With it,cua do snapshotreturns an AI-annotated screen with element coordinates. Without it, usescreenshotand read the image yourself.
Workflow
Look → Act → Verify — repeat until done, then share:
cua do screenshot # look
cua do click 450 280 # act
cua do screenshot # verify
cua trajectory share # share replay link with user
Re-screenshot after every UI change — coordinates go stale when the screen changes.
Scenarios
Click a button
cua do screenshot
cua do click 450 280
cua do screenshot
Fill a form
cua do screenshot
cua do click 400 200 && cua do type "Jane Doe"
cua do key tab && cua do type "[email protected]"
cua do key tab && cua do type "SecureP@ss123"
cua do click 400 500
cua do screenshot
File upload dialog
cua do click 350 400 # "Choose File"
cua do type "/home/user/report.pdf"
cua do key enter
cua do screenshot
Zoom in for precision clicks (host or small targets)
When clicking small or dense UI elements — especially on the host machine — zoom into the target window first. Coordinates become window-relative and screenshots show only that window, giving you higher effective resolution.
cua do zoom "Google Chrome" # crop to Chrome window; coords are now window-relative
cua do screenshot # zoomed view — easier to locate small elements
cua do click 112 44 # precise click on a small tab or button
cua do screenshot # verify
cua do unzoom # restore full-screen coords when done
cua do screenshot # back to full desktop view
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 162 lines · 68 tokens per session scan A 18d28e0a1ca4
gui-automation is a skill published in the GitHub repository trycua/cua (22,200 stars, last pushed today), licensed MIT. It adds 68 tokens to every session and 1,423 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
browser-test-tauri-fe
Test, screenshot, and debug a Tauri or Vite frontend in a real browser using Playwright or Chrome DevTools MCP. Use when asked to run/screenshot a desktop-app or web UI, verify a frontend change visually, debug rendering without the native build, or when the native webview (macOS WKWebView) can't be driven by…
mirroir-onboard
Onboard a consumer web app to mirroir's .mirroir/ dotfile by EXPLORING the running app (chrome-devtools-mcp) — derive real selectors from the accessibility tree, exercise each surface's primary action, emit the .mirroir/ tree, and validate by LIVE REPLAY with a self-heal loop. Reject shallow "page renders" coverage.
walkthrough
Generate a hands-on browser walkthrough of a PR's user-facing changes to exercise before review; --publish posts the final version to the PR for QA.
webapp-testing
Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.
playwright-cli
Translate natural-language browser automation requests into exact playwright-cli commands for interactive web testing and debugging. Use when requests involve opening/navigating pages, interacting with elements, capturing snapshots/screenshots/PDFs, using tabs, inspecting console/network, mocking routes, managing…
playwright-cli
Automate browser interactions, test web pages and work with Playwright tests.