manushi4

manushi4 / Screenhand

Public

Give AI eyes and hands on your desktop. Open-source MCP server for desktop automation — screenshots, UI control, browser automation, OCR. Works with Claude, Cursor, and any MCP client. macOS + Windows.

screenhand.com accessibility ai-agent browser-automation claude computer-use
12
0
100% credibility
Found Mar 06, 2026 at 12 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

ScreenHand is an open-source desktop automation bridge that enables AI assistants to interact with macOS and Windows applications using native accessibility APIs, OCR screenshots, keyboard/mouse simulation, and Chrome browser control via an MCP server.

How It Works

1
💡 Discover ScreenHand

You hear about ScreenHand, a helpful tool that lets your AI friend like Claude see and control apps on your computer screen.

2
📥 Get it set up

Download and install it quickly on your Mac or Windows, just like any app.

3
🔓 Allow screen access

Give it permission to see and interact with your screen, a one-time simple step in your settings.

4
🔗 Link to your AI

Connect it to Claude or your favorite AI helper so they can now understand and use your desktop.

5
🖱️ Watch AI take action

Tell your AI what to do, like 'click the send button' or 'type my name', and it handles everything smoothly.

6
🚀 Automate daily tasks

Your AI now reads buttons, fills forms, navigates apps, and gets work done faster without you lifting a finger.

🎉 Effortless control achieved

Enjoy a smarter desktop where AI manages apps reliably, saving you time every day.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 12 to 12 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is Screenhand?

Screenhand gives AI agents eyes and hands on your desktop through an open-source TypeScript MCP server. It delivers screenshots with OCR, native UI inspection/control via accessibility APIs, keyboard/mouse input, and Chrome automation over CDP—all on macOS and Windows with one interface. Developers plug it into Claude, Cursor, or any MCP client to let AI see screens, click elements, type text, and chain app workflows without cloud dependencies.

Why is it gaining traction?

Unlike screenshot-heavy tools like Anthropic Computer Use or OpenClaw, Screenhand hits ~50ms native actions with zero extra AI calls, plus OCR fallbacks and stealth CDP for bot detection. Built-in memory auto-learns strategies from sessions, recalls fixes for errors, and exports playbooks for sites like Devpost. Cross-platform parity and slash commands (/screenshot, /debug-ui) hook devs fast.

Who should use this?

AI agent builders wiring Claude/Cursor for desktop tasks, like giving GitHub Copilot context from local apps or automating browser flows. QA folks scripting UI tests across apps. Frontend devs tired of manual form filling who want to give eyes to dodomeki-like prototypes or give GitHub repo access via UI automation.

Verdict

Promising MCP bridge at 1.0% credibility (12 stars), with strong docs, quickstart configs, and solid test coverage—but still early, so expect tweaks for edge apps. Try it if you need local desktop control; contribute to boost Windows support.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.