computer_use

computer_use is a skill for Claude Code, Codex from IceWhaleTech/ZimaOS-Blue. It costs 99 tokens per session (1,694 once invoked), scanned A, original, MIT.

A set of instructions for inspecting and controlling a live desktop app, native window, or browser interface through accessibility controls.

In plain words
What is it for?
Use it to focus an app, inspect its controls, click buttons, change settings, select items, type text, scroll, or send a message.
Why use it?
It helps an agent work with visible applications when editing files or using a normal command line is not enough.

Skill for Claude CodeCodex

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

Good fit Use it to focus an app, inspect its controls, click buttons, change settings, select items, type text, scroll, or send a message.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/icewhaletech/zimaos-blue/computer_use
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 IceWhaleTech/ZimaOS-Blue --skill computer_use
Clone the repo
git clone --depth 1 https://github.com/IceWhaleTech/ZimaOS-Blue

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 computer_use

README.md
[![agentmods](https://agentmods.dev/badge/skills/icewhaletech/zimaos-blue/computer_use.svg)](https://agentmods.dev/skills/icewhaletech/zimaos-blue/computer_use)
Your own site
<a href="https://agentmods.dev/skills/icewhaletech/zimaos-blue/computer_use"><img src="https://agentmods.dev/badge/skills/icewhaletech/zimaos-blue/computer_use.svg" alt="Measured on agentmods" height="20"></a>
Per session 99 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,694 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 $0.00099 $0.01694
Opus 5 $0.00049 $0.00847
Sonnet 5 $0.00020 $0.00339
Haiku 4.5 $0.00010 $0.00169

Measured 7d ago against content hash ce9f0f25224f, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-07, from the pricing page.

Security

Grade A, and why

computer_use 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.

assets/skills/computer_use/SKILL.md · 158 lines

How it starts

The opening of the file, as written. The whole thing — 158 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Computer Use

Use the built-in computer_use tool for live UI work on desktop apps, native windows, and accessibility-backed browser flows.

Setup

No external CLI is required.

For host-window actions:

  • macOS: Accessibility permission may be required.
  • Windows: UI Automation / WinEvent-backed inspection must be available.

If the user only wants to edit a workspace artifact such as .docx, .xlsx, .pptx, or .pdf, use the native document tools instead of driving the app UI.


Task Routing

User Intent Action
Focus the right app or window blue computer_use action=focus ...
Inspect interactive host UI before acting blue computer_use action=snapshot_interactive ...
Read the full accessibility tree for debugging blue computer_use action=snapshot ...
Send a chat message in a known conversation blue computer_use action=message ...
Switch to a conversation/list row/item without typing blue computer_use action=select ...
Type into the likely composer/input without exposing refs blue computer_use action=type ...
Click a named control or toggle a named setting blue computer_use action=click ... / toggle ...
Scroll, press keys, or capture evidence blue computer_use action=scroll ..., key ..., screenshot ...
Run a multi-step CUA desktop task blue computer_use action=task goal="..."
Use CUA action aliases open_app, input_text, Click, RightSingle, move_mouse, scroll_up/down, Hotkey, multi_Hotkey, record_info, done
User goal is still ambiguous and you truly need low-level control blue computer_use action=act params...

Command Usage

Fast Path

blue computer_use action=message app_name="Feishu,Lark" conversation="Orca" value="hello"
blue computer_use action=type app_name="Feishu,Lark" conversation="Orca" value="draft only"
blue computer_use action=select app_name="Feishu,Lark" conversation="Orca"
blue computer_use action=click window_title="Settings" control="Open Network"
blue computer_use action=toggle window_title="Settings" setting="Enable notifications"

Read the full file on GitHub · 158 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 · 158 lines · 99 tokens per session scan A ce9f0f25224f

Subscribe to this mod's changes

computer_use is a skill published in the GitHub repository IceWhaleTech/ZimaOS-Blue (25 stars, last pushed 4mo ago), licensed MIT. It adds 99 tokens to every session and 1,694 once invoked, about $0.0005 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

credential-setup-with-computer-use

Guides n8n credential setup through Computer Use browser tools. Use when a user needs OAuth apps, API keys, client IDs, client secrets, or other credential values from an external service console.

n8n-io/n8n · 48 tokens

defuddle

Extract clean markdown content from web pages using Defuddle CLI, removing clutter and navigation to save tokens. Use instead of WebFetch when the user provides a URL to read or analyze, for online documentation, articles, blog posts, or any standard web page. Do NOT use for URLs ending in .md — those are already…

kepano/obsidian-skills · 74 tokens

browser-harness-skill

Drive the user's real Chrome over raw CDP by writing Python against browser-harness helpers - screenshot, coordinate clicks, JS evaluation, form fill, tabs. For tasks needing full freedom or the user's own logins.

zeenie-ai/OpenCompany · 49 tokens

browser-skill

Interactive browser automation - navigate, click, type, fill forms, take screenshots, get accessibility snapshots. Supports system Chrome/Edge via auto-detection.

zeenie-ai/OpenCompany · 34 tokens

nearby-places-skill

Search for nearby places like restaurants, cafes, stores, and services using Google Places API. Find places by type and location.

zeenie-ai/OpenCompany · 32 tokens

unbrowse

One-call web access for agents with cache-first API replay and browser capture on misses. Unbrowse passively learns first-party route DAGs while browsing, independently validates replay, and keeps remote sharing consented and fail-closed. Prefer it over WebFetch, curl, and browser loops.

unbrowse-ai/unbrowse · 60 tokens