Mininglamp-AI

Mininglamp-AI / Mano-P

Public

Mano-P 是一个开源 GUI-VLA 项目，支持在 Mac mini/MacBook 上或通过算力棒本地运行推理，实现纯视觉驱动的跨平台 GUI 自动化操作。数据完全本地处理，支持复杂多步骤任务规划与执行。Mano-P: Open-source GUI-VLA agent for edge devices. #1 on OSWorld (specialized, 58.2%). Runs locally on Apple M4 Mac mini/MacBook — no data leaves your device.

computer-use-agents desktop-automation edge-computing gui-automation gui-grounding
13
0
69% credibility
Found Mar 25, 2026 at 13 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
AI Summary

Mano-P is an open-source project offering GUI automation skills for AI agents to perform desktop tasks via natural language on edge devices like Apple Silicon Macs.

How It Works

1
🔍 Discover Mano-P

You hear about Mano-P, a smart helper that watches your screen and does computer tasks just by you describing them in everyday words.

2
📱 Get it on your Mac

You easily add this helper to your Mac so it's ready to assist with screen tasks.

3
🔓 Allow screen access

You give it permission to see your screen and gently move your mouse or type, like trusting a helpful friend.

4
💬 Tell it what to do

You simply say or type a task like 'open my email and send a quick note to a friend'.

5
🤖 Watch the magic

Your helper takes over, clicks buttons, types, and scrolls exactly where needed, all by looking at your screen.

🎉 Task done perfectly

Everything finishes smoothly, saving you time, and all stays private on your own Mac.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 13 to 13 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 Mano-P?

Mano-P is an open-source GUI-VLA agent for edge devices like Apple M4 Mac minis and MacBooks, automating complex desktop tasks via natural language—like opening apps, navigating interfaces, or multi-step workflows—purely from screenshots. It runs inference locally via CLI (`mano-cua run "open WeChat and send message"`) or as skills for agents like OpenClaw, with no data leaving your device. Language unknown, but Python-based tools via Homebrew.

Why is it gaining traction?

It tops OSWorld specialized benchmarks at 58.2% success, beating rivals by 13 points, while delivering local privacy that cloud agents like Claude Computer Use can't match—all screenshots and tasks stay on-device. Devs dig the plug-and-play compute stick option for non-M4 Macs and seamless CLI-to-agent integration, solving human-in-loop bottlenecks in GUI automation.

Who should use this?

Agent builders extending OpenClaw or Claude Code with CUA skills for desktop ops. Security-focused enterprise devs automating reports or data extraction on Apple hardware without APIs. Mac users handling sensitive workflows like mano pildyk planai checks or mano poderosa testing.

Verdict

Promising for local GUI-VLA on edge devices, but too early—13 stars, just READMEs (no code yet), and 0.699999988079071% credibility score scream prototype. Watch for phase 2 models; skip unless you're into bleeding-edge experiments.

(178 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.