Computer Use Windows

Computer Use Windows is a skill for Claude Code, Codex from fanfan-de/anybox. It costs 22 tokens per session (1,071 once invoked), scanned A, original, MIT.

A tool for controlling approved windows in Windows desktop applications through Anybox. It is intended for operating apps through their visible windows.

In plain words
What is it for?
Use it to operate supported Windows desktop applications during an automated task.
Why use it?
It helps an agent interact with Windows programs when no structured integration is available.

Skill for Claude CodeCodex

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

Good fit Use it to operate supported Windows desktop applications during an automated task.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/fanfan-de/anybox/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 fanfan-de/anybox --skill computer-use
Clone the repo
git clone --depth 1 https://github.com/fanfan-de/anybox

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 Windows

README.md
[![agentmods](https://agentmods.dev/badge/skills/fanfan-de/anybox/computer-use/github.svg)](https://agentmods.dev/skills/fanfan-de/anybox/computer-use)
Your own site
<a href="https://agentmods.dev/skills/fanfan-de/anybox/computer-use"><img src="https://agentmods.dev/badge/skills/fanfan-de/anybox/computer-use/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for Computer Use Windows

Your own site · 80×15
<a href="https://agentmods.dev/skills/fanfan-de/anybox/computer-use"><img src="https://agentmods.dev/badge/skills/fanfan-de/anybox/computer-use.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 22 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,071 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. Third-party audits
  • NVIDIA SkillSpector warn 7 Sept 2026
SkillSpector: 1 finding, up to medium

These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →

  • medium Excessive Agency · line 66
    Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
    Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
How audits are shown
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 $0.00022 $0.01071
Opus 5 $0.00011 $0.00535
Sonnet 5 $0.00004 $0.00214
Haiku 4.5 $0.00002 $0.00107

Measured 8d ago against content hash c26d8642848c, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

Computer Use Windows 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 8d 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.

plugins/Anybox-Plugins/computer-use-windows/skills/computer-use/SKILL.md · 71 lines

How it starts

The opening of the file, as written. The whole thing — 71 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Computer Use Windows

Use this skill when the user asks you to operate a Windows desktop application and no safer structured integration is available.

Initialize once per Anybox session

Use the Anybox js Node REPL tool. Resolve this installed skill's plugin root from the absolute SKILL.md path shown in the skill catalog, then import scripts/computer-use-client.mjs by absolute path:

if (!globalThis.sky) {
  const { setupComputerUseRuntime } = await import(
    "<absolute-plugin-root>/scripts/computer-use-client.mjs"
  );
  await setupComputerUseRuntime({ globals: globalThis });
}
return { ready: Boolean(globalThis.sky), target: globalThis.sky?.target };

Do not spawn the native helper, open its named pipe, or implement another protocol client. Importing the plugin client is the only supported initialization path; it owns the Computer Use runtime and helper lifecycle.

The Node kernel persists only within the current Anybox session. A new session or js_reset terminates the old process, so run this initialization again whenever globalThis.sky is absent.

Node REPL output contract

The js tool executes submitted code as an async function body. It does not echo the final expression like an interactive console. A bare expression such as await sky.list_apps() still runs, but its tool result is null unless the code explicitly returns or writes the value.

Always use return for any value the model must inspect. Keep reusable handles on globalThis so later calls retain the exact Window objects:

globalThis.computerUseApps = await sky.list_apps();
return globalThis.computerUseApps;

Prefer return over nodeRepl.write(...) for structured results. If a call is intentionally side-effect-only, still return a short verification object instead of relying on an empty result.

Operating loop

Keep the sky object and returned Window objects in the current session's persistent REPL. Never expect those handles to survive a session change or hard reset.

Read the full file on GitHub · 71 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. 8d ago First seen · 71 lines · 22 tokens per session scan A c26d8642848c

Subscribe to this mod's changes

Computer Use Windows is a skill published in the GitHub repository fanfan-de/anybox (57 stars, last pushed 29d ago), licensed MIT. It adds 22 tokens to every session and 1,071 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 2026-09-03.