Give AI full desktop and browser control
Author built a tool giving agents full desktop control plus a browser debugger extension, after OpenAI computer use refused menu bar interaction; used daily with Claude and Codex. ·
- What they did
- The author built a tool that gives an AI agent full desktop control, plus a browser debugger extension for working inside the browser. They made it because OpenAI's computer use was limited and refused to interact with the menu bar or task tray. The tool is MIT licensed and connects to Claude and Codex.
- How it went
- The author says they use it daily with Claude and Codex. The comment gives no numbers, benchmarks, or details on failures.
- Worth knowing
- The source is a brief comment with no setup steps or costs; check the linked machinecontrol.dev site for how to install and configure it.
Try it yourself with Claude
Help me build a small tool that lets an AI agent control my [Mac/Windows/Linux] desktop, including menu bar and system UI clicks, and pairs with a browser extension for debugging pages. Start by listing the pieces we need, then build the simplest working version that can open [an app], click a menu bar item, and report what happened. Pause and show me the plan before running anything that changes system settings or deletes files.
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
Source: HN · Nov 17, 2026
Related use cases
Build a C compiler with a team of parallel agents
Anthropic's engineering write-up on tasking Opus 4.6 agent teams with building a C compiler.
Schedule 60 short videos in YouTube Studio
Claude via the Chrome extension scheduled 60 generated short videos in YouTube Studio after suggesting the task itself.
Publish one video to four Chinese platforms at once
A Codex skill generates titles and tags, then sub-agents fill upload forms on Xiaohongshu, Douyin, Bilibili and WeChat Channels in parallel; the author confirms each post.
Submitted by from search gap: 抖音
Review Linux kernel patches
Google engineers launched Sashiko, an agentic AI code reviewer for the Linux kernel.
Custom agent
Diagnose Kubernetes alerts with MCP servers
An n8n agent triggered by Alertmanager queries Kubernetes, Grafana and other MCP servers and posts a diagnostic report with root-cause hypotheses in a Mattermost thread.
Auto-pull proven skills for Cursor on an analytics dashboard
On their analytics dashboard, Cursor's agent pulled in skills for event-store design, instrumentation and API security, adding rate limiting and input validation.
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.