Mmcp.market

computer-use skill

by Razshy·Razshy/Wiggle·91 stars

Read this skill before the first step of any request to do something in an app on the person's own computer (Notes, Finder, System Settings, any desktop app), to look at their screen, or for \"computer use\". Computer use (desktop control) lets Claude take screenshots of the person's desktop and control it with clicks, typing and scrolling through the Claude desktop app; its tools are named mcp__computer-use__* when the session runs in the desktop app and mcp__remote-devices__computer_* when a cloud session is linked to the person's computer; before a conversation is linked there may be no such tools, only an enable__mcp__remote-devices__Claude_Browser tool, which links it. It covers getting linked, picking the right tool, the access flow, and what to do when computer use is off or out of reach. It is not for websites, which go through Claude in Chrome or the built-in browser and their own skills.

A100/100content scan

Is the computer-use skill safe?

Clean: nothing in its files matched our rules. We read 2 files in the folder on 2026-09-28.

No findings.

Install the computer-use skill

A skill is a folder. Copy it into your agent's skills folder and the agent loads it when the task matches its description. 

git clone --depth 1 https://github.com/Razshy/Wiggle.git /tmp/Wiggle
mkdir -p ~/.claude/skills
cp -r /tmp/Wiggle/mnt-skills/examples/computer-use ~/.claude/skills/computer-use
available in every project

In the Claude apps, zip the folder and upload it from the Skills settings. The folder on GitHub

The instructions your agent would load

SKILL.md as published, without the frontmatter. Read it on GitHub

Computer use (desktop control)

Computer use lets Claude take screenshots of the person's desktop and control it with mouse clicks, keyboard input, and scrolling, through the Claude desktop app. Its tools are named mcpcomputer-use (for example requestaccess, screenshot) when the session runs inside the desktop app, and mcpremote-devicescomputer (for example computerrequestaccess, computerscreenshot) when the session runs in the cloud and is linked to the person's computer. If they are deferred behind ToolSearch, Claude loads them all in one call (query "computer", a generous maxresults such as 40) rather than one by one.

Getting linked from a chat

What Claude does first depends on which tools are present (loaded or deferred):

  1. Computer-use tools are present. The conversation is already linked; Claude goes straight to the access flow below.
  2. No computer-use tools, but an enablemcpremote-devicesClaudeBrowser tool is present. The conversation is not linked to the person's computer yet. Claude:
  3. Calls that enable tool first, even though its description speaks only of the browser, with the computer task in its task field, and tells the person in one line that it is linking to their computer through the Claude desktop app (the prompt they may see mentions its browser) and that Computer use needs to be turned on under Settings → Desktop app → Computer use there. If that call is declined or answers that the browser could not be used, the computer was not linked either; Claude says so in terms of the computer and does not retry unless asked.
  4. Checks for the mcpremote-devicescomputer tools once it has run: linking the Claude desktop app for its browser links it for computer use too, but the computer tools appear only when computer use is available on that computer, usually only once it is turned on in the desktop app (it is off by default; the last section covers when they do not appear). In some organizations each computer_* action may also ask the person for approval.
  5. Treats the note that follows as being about the built-in browser: it says to use the browser's tools, and may instead say the browser is not linked. When the request was for a desktop app rather than a website, that instruction points at the wrong tool and says nothing about computer use; Claude checks for computer_* tools, carries the request out with them starting with the access flow, uses the browser only for any website part of the request, and otherwise follows the last section.

While this enable tool is available Claude does not say it cannot use the person's computer, and it does not say it can until a computer_* call has succeeded.

  1. Neither. Computer use is not available in this conversation as it stands, and Claude says so plainly without diagnosing why: in general Claude can use apps on a computer where the Claude desktop app is installed and signed in with Computer use turned on, and on claude.ai in a web browser, if the composer's + menu offers Devices, the person can pick their computer there before sending a new message. If other mcpremote-devices tools are present, the last section applies instead. Claude does not pretend to act.

Separate filesystems

Computer-use actions (clicks, typing, clipboard writes) happen on the person's real computer, a different system from wherever Claude's own shell and files are, when it has them. Files Claude creates on its side do not exist on the person's machine, so a command or file path Claude puts in the person's clipboard or types into one of their apps must exist on their computer; and a path the person mentions (~/Documents/..., C:\Users\...) is on their computer, not in Claude's working directory.

Pick the right tool for the app

Each tier trades speed and precision against coverage:

  1. Dedicated connector for the app. If the task is in an app that has its own MCP connector (Slack, Gmail, Calendar, Linear, etc.) and it is connected, Claude uses it. API-backed tools are fast and precise.
  2. A browser (Claude in Chrome or the built-in browser; the chrome-browser and built-in-browser skills describe their tools). If the target is a web app with no dedicated connector, Claude uses the browser tools: DOM-aware, much faster than clicking pixels. Computer use cannot operate web browsers (browser windows are view-only to it, for safety), so if no browser is connected Claude asks the person to set one up rather than falling through to computer use.
  3. Computer use, for native desktop apps (Maps, Notes, Finder, Photos, System Settings, any third-party native app) and cross-app workflows. Computer use is the right tool here; Claude does not decline a native-app task just because there is no dedicated connector for it.

This is about what is available, not error handling: if a connector tool errors, Claude debugs or reports it rather than silently retrying through a slower tier.

Look before asserting

If the person asks about app state (what is open, what is connected, what an app can do), Claude takes a screenshot and checks before answering, rather than answering from memory; the person's setup or app version may differ from what Claude expects. A claim that an app does not support an action should be grounded in what Claude just saw on screen, not general knowledge; listgrantedapplications or a fresh screenshot is cheaper than a wrong assertion about what is running. Likewise Claude does not describe the screen or call the computer linked until a computer-use call has succeeded.

Access flow

Before any computer-use action Claude must request access to the applications it needs; the person approves them explicitly, and Claude may need to ask again mid-task if it discovers it needs another application. With the mcpcomputer-use tools that is one requestaccess call listing the applications. With the mcpremote-devicescomputer tools it is two calls: computerresolveaccess with the app names, then computerrequestaccess with the entries it returned, verbatim, and a one-sentence reason that explains the task. Claude waits for the person's answer rather than working around it.

When computer use is off or out of reach

With the mcpremote-devicescomputer_* tools, computer use works only while the Claude desktop app is open and online on the person's computer with Computer use turned on in its settings, where it is off by default. Claude matches what it sees to one of these, never calls any of them the browser being unlinked or disconnected, and never claims an action happened:

  • mcpremote-devices tools are present but no computer tools among them, or a computer call answers that computer use is turned off or not turned on yet. The setting is most likely off. If the reply says to call computerrequestaccess with only a reason to show the person a prompt, Claude does that once; in a conversation linked through the enable tool above that prompt may not be shown, so if nothing changes Claude does not wait for it. Claude tells the person to turn on Computer use in the Claude desktop app (Settings → Desktop app → Computer use) with the app up to date and open, tries again once they say it is on (the tools can take a few minutes to appear), and if they still do not appear says so and suggests a new chat.
  • A call answers that computer use is not available or not supported on this device, or not allowed for the organization. Claude relays that plainly and does not promise a setting will fix it.
  • The calls cannot reach the desktop app at all (connection errors or no response). Claude says the computer looks closed, asleep or offline, and continues with what it can do here.

More skills from Razshy/Wiggle

  • Aalgorithmic-artCreating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations.
  • Abenepass-reimbursementSubmit expense reimbursements through Benepass (app.getbenepass.com). For users whose employer uses Benepass as their benefits platform. Handles login, benefit selection, form filling, receipt upload, and submission. Requires browser/computer-use capabilities.
  • Abrand-guidelinesApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.
  • Abuilt-in-browserRead this skill before the first step that uses the built-in browser, the browser pane inside the Claude desktop app (also called the in-app browser, the browser pane, Claude's browser, or \"your own browser\"), whose tools are named mcp__Claude_Browser__* when the session runs in the desktop app and mcp__remote-devices__Claude_Browser__* when a cloud session is linked to the person's computer; before those tools are turned on there may be a single enable__mcp__remote-devices__Claude_Browser tool instead. It covers the pane's persistent sign-ins, tabs and preview_start, reading pages as text, site approvals, what the pane cannot open, and what to do when it cannot be reached. It is not for Claude in Chrome (mcp__claude-in-chrome__* tools), which has its own skill, and it does not decide which browser to use.
  • Acall-to-bookMake a phone call to book an appointment or reservation. Checks calendar first, gets explicit consent before dialing, discloses AI identity on the call, and adds the booking to calendar when done.
  • Acancel-unsubscribeCancel a subscription or unsubscribe from a service. Works from a description, a pasted charge line, a URL, or a photo/screenshot. Can also audit a full statement for recurring charges and cancel several at once. Finds the right contact method and handles the cancellation — including phone calls.
  • Acanvas-designCreate beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations.
  • Achrome-browserRead this skill before the first step that uses Claude in Chrome, the browser extension whose tools are named mcp__claude-in-chrome__* (also called Chrome, the browser extension, or the external browser) and which acts in the person's real Chrome with their own sign-ins; before those tools are turned on there may be a single enable__mcp__claude-in-chrome tool instead. It covers loading the tools in one ToolSearch call, checking the person's open tabs and working in a new tab, site permissions, GIF recordings, console logs, dialogs to avoid, and when to stop and ask. It is not for the built-in browser (mcp__Claude_Browser__* or mcp__remote-devices__Claude_Browser__* tools), which has its own skill, and it does not decide which browser to use.
  • Adeep-researchUse this skill when the user's prompt requires (1) researching a topic across multiple sources, comparing options or alternatives, analyzing trends or history, understanding markets or industries, or reviewing literature or studies and (2) synthesizing that research into a comprehensive, narrative report. If you're planning to search the web or internal knowledge bases, consider using this skill. This skill coordinates research subagents, so use it only when you have a tool for spawning subagents (the Agent or Task tool); otherwise, research the question directly.
  • Adoc-coauthoringGuide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.
  • AdocxUse this skill whenever the user wants to create, read, edit, or manipulate Word documents (.docx) or Word templates (.dotx). Triggers include: any mention of 'Word doc', 'word document', '.docx', '.dotx', or requests to produce professional documents with formatting like tables of contents, page numbers, or letterheads. Also use when extracting or reorganizing content from .docx or .dotx files, inserting or replacing images in documents, find-and-replace in Word files, working with tracked changes or comments, or converting content into a polished Word document. If the user asks for a 'report', 'memo', 'letter', 'template', or similar deliverable as a Word or .docx file (to download, email or print), use this skill. However, if they ask for a document, page, report, memo, or notes WITHOUT naming a file format and the session offers a dedicated document or page skill or connector, use that instead. Do NOT use for PDFs, spreadsheets, Google Docs, or coding unrelated to document generation.
  • Aevent-planningHelp plan an event — from a birthday dinner to a wedding. Scales to the size of the occasion. Handles venue research, guest lists, timelines, vendors, and budgets.

All agent skills → · MCP servers