CtrlK
BlogDocsLog inGet started
Tessl Logo

computer-use

Use this skill when the user wants to automate desktop or browser tasks — clicking, typing, screenshotting, running GUI apps, or filling web forms.

67

Quality

81%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Low

Low-risk findings worth noting

SKILL.md
Quality
Evals
Security

Computer Use

Overview

Provides Claude with tools to interact with a computer: take screenshots, move the mouse, click, type text, and execute bash commands. Based on the Anthropic computer-use-demo quickstart with Prowlr integration.

Tools Available

# Screenshot — capture current screen state
{
  "name": "computer",
  "input": {
    "action": "screenshot"
  }
}

# Click at coordinates
{
  "name": "computer",
  "input": {
    "action": "left_click",
    "coordinate": [x, y]
  }
}

# Type text
{
  "name": "computer",
  "input": {
    "action": "type",
    "text": "Hello, world!"
  }
}

# Run bash command
{
  "name": "bash",
  "input": {
    "command": "ls -la ~/Desktop"
  }
}

Usage Pattern

import anthropic

client = anthropic.Anthropic()

# Start with a screenshot to understand current state
tools = [
    {"type": "computer_20241022", "name": "computer", "display_width_px": 1920, "display_height_px": 1080},
    {"type": "bash_20241022", "name": "bash"},
    {"type": "text_editor_20241022", "name": "str_replace_editor"},
]

response = client.beta.messages.create(
    model="claude-opus-4-6",
    max_tokens=4096,
    tools=tools,
    messages=[{
        "role": "user",
        "content": "Open the browser, navigate to example.com, and take a screenshot."
    }],
    betas=["computer-use-2024-10-22"],
)

Setup Requirements

# Docker container with display server recommended
docker run -it \
  -e DISPLAY=:1 \
  -v /tmp/.X11-unix:/tmp/.X11-unix \
  anthropic/computer-use-demo

# Or use with Xvfb on Linux
Xvfb :1 -screen 0 1920x1080x24 &
export DISPLAY=:1

Safety Notes

  • Run in a sandbox/container — Claude will have real control of the system
  • Review actions before executing in production environments
  • Use --read-only filesystem mounts where possible
  • This skill has MEDIUM risk: real system access, network access possible

Quick Reference

ActionInput
Screenshot{"action": "screenshot"}
Click{"action": "left_click", "coordinate": [x, y]}
Type{"action": "type", "text": "..."}
Key press{"action": "key", "text": "ctrl+c"}
Scroll{"action": "scroll", "coordinate": [x, y], "direction": "down"}
Bash{"command": "..."}
Repository
ProwlrBot/prowlr-marketplace
Last updated
First committed

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.