gui-agent

gui-agent is a skill for Claude Code, Codex from THU-SAGE/syll. It costs 13 tokens per session (697 once invoked), scanned A, original, MIT.

A tool for controlling desktop applications through screenshots and automated mouse-and-keyboard actions. It repeatedly observes the screen, chooses an action, performs it, and checks the result.

In plain words
What is it for?
Use it to open applications, click controls, type text, scroll, and complete other GUI workflows. It can also help with tasks that require repeated visual verification.
Why use it?
It helps complete tasks in applications that do not offer a direct programming interface. Screen checks and safety rules reduce the chance of acting on an unexpected state.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit Use it to open applications, click controls, type text, scroll, and complete other GUI workflows. It can also help with tasks that require repeated visual verification.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/thu-sage/syll/gui-agent
Install

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.

Any agent
npx skills add THU-SAGE/syll --skill gui-agent
Clone the repo
git clone --depth 1 https://github.com/THU-SAGE/syll

Made for: Claude Code, Codex.

Wrote 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.

agentmods badge for gui-agent

README.md
[![agentmods](https://agentmods.dev/badge/skills/thu-sage/syll/gui-agent.svg)](https://agentmods.dev/skills/thu-sage/syll/gui-agent)
Your own site
<a href="https://agentmods.dev/skills/thu-sage/syll/gui-agent"><img src="https://agentmods.dev/badge/skills/thu-sage/syll/gui-agent.svg" alt="Measured on agentmods" height="20"></a>
Per session 13 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 697 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00013 $0.00697
Opus 5 $0.00006 $0.00349
Sonnet 5 $0.00003 $0.00139
Haiku 4.5 $0.00001 $0.00070

Measured 7d ago against content hash 1e223833066a, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-07, from the pricing page.

Security

Grade A, and why

gui-agent 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 7d 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.

syll/skills/gui-agent/SKILL.md · 86 lines

How it starts

The opening of the file, as written. The whole thing — 86 lines — stays where its author put it; the contents beside it link to each section on GitHub.

GUI Agent

You can interact with desktop applications using the gui_action tool.

How It Works

  1. The tool captures a screenshot of the current screen
  2. Sends it to the UI-TARS vision model for analysis
  3. UI-TARS returns a thought + action (click, type, scroll, etc.)
  4. The action is executed via pyautogui
  5. Steps repeat until the task is complete or max steps reached

Usage Protocol

When the user asks you to perform a GUI task:

  1. Understand the goal - Break the task into clear steps
  2. Call gui_action with a clear instruction
  3. Review the result - Check the returned screenshot and status
  4. Iterate if needed - Call again with refined instructions

Example

User: "Open Chrome and search for 'Syll'"

You should call:

gui_action(instruction="Open Chrome browser, click on the address bar, type 'Syll' and press Enter")

Safety Rules

  1. Never perform destructive actions without user confirmation:

    • Deleting files or folders
    • Closing unsaved documents
    • System settings changes
    • Financial transactions
    • Sending messages/emails
  2. Always verify the current screen state before acting

  3. Stop immediately if the screen shows unexpected content (login pages with sensitive data, etc.)

  4. Report clearly what actions were taken and their results

Supported Actions

Action Description Example
click Left click at position click(point='<point>500 300</point>')
right_click Right click right_click(point='<point>500 300</point>')
double_click Double click double_click(point='<point>500 300</point>')
drag Drag from A to B drag(start='<point>100 100</point>', end='<point>200 200</point>')
type Type text type(content='hello world')
hotkey Key combination hotkey(key='ctrl+c')
scroll Scroll up/down scroll(point='<point>500 300</point>', direction='down', amount=3)
wait Pause wait(seconds=2)
finished Task complete finished(content='Done')

Read the full file on GitHub · 86 lines

Changes

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.

  1. 7d ago First seen · 86 lines · 13 tokens per session scan A 1e223833066a

Subscribe to this mod's changes

gui-agent is a skill published in the GitHub repository THU-SAGE/syll (303 stars, last pushed 3mo ago), licensed MIT. It adds 13 tokens to every session and 697 once invoked, about $0.0001 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.

Related

Other skills, from other repositories

opencli-sitemap-author

Use when creating or maintaining OpenCLI site sitemaps: agent-facing navigation, page-state, action, workflow, API-reference, pitfall, and fallback knowledge for a website. Use after browser exploration discovers durable site context, when a sitemap is stale, or when promoting local site knowledge into the repo.

jackwener/OpenCLI · 67 tokens

debug-optimize-lcp

Guides debugging and optimizing Largest Contentful Paint (LCP) using Chrome DevTools MCP tools. Use this skill whenever the user asks about LCP performance, slow page loads, Core Web Vitals optimization, or wants to understand why their page's main content takes too long to appear. Also use when the user mentions…

ChromeDevTools/chrome-devtools-mcp · 99 tokens

interactive-login

How to complete browser/interactive logins (aws / gh / glab / gcloud). The platform backgrounds the login poller so it survives the human's browser round-trip — and when that does NOT work.

yc-software/qm · 46 tokens

pinchtab-mcp

Use this skill when a task requires browser automation through PinchTab's MCP server connected to a remote browser instance. Covers navigation, element interaction, data extraction, form filling, multi-step flows, and session management via MCP tools.

pinchtab/pinchtab · 52 tokens

azure-messaging-webpubsub-java

Build real-time web applications with Azure Web PubSub SDK for Java. Use when implementing WebSocket-based messaging, live updates, chat applications, or server-to-client push notifications.

microsoft/skills · 43 tokens

google-safe-browsing

Prevent and fix Google Safe Browsing "Dangerous site" flags. Use when launching a public web app, buying/picking a domain, building a login or signup page, or when any site shows a red "Dangerous site" / "Deceptive site" warning in Chrome, Brave, Safari, Firefox, or Edge. Triggers on "dangerous site", "deceptive…

davidondrej/skills · 105 tokens