Computer – Build Your Manus AI Agent with an OSS macOS Sandbox
We just open-sourced Computer, a Computer-Use Interface (CUI) framework that enables AI agents to interact with isolated macOS and Linux sandboxes, with near-native performance on Apple Silicon. Computer provides a PyAutoGUI-compatible interface that can be plugged into any AI agent system (OpenAI Agents SDK , Langchain, CrewAI, AutoGen, etc.). Why Computer? As CUA AI agents become more capable, they need secure environments to operate in. Computer solves this with: • Isolation: Run agents in sandboxes completely separate from your host system. • Reliability: Create reproducible environments…
In plain words
Computer is an open-source framework that lets AI agents interact with isolated macOS and Linux sandboxes with near-native performance on Apple Silicon. It provides a PyAutoGUI-compatible interface compatible with OpenAI Agents SDK, Langchain, CrewAI, and AutoGen. The tool emphasizes security through sandboxing, allowing agents to operate in completely isolated environments separate from the host system, while maintaining reliability and giving users control to monitor and terminate workflows.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
We just open-sourced Computer, a Computer-Use Interface (CUI) framework that enables AI agents to interact with isolated macOS and Linux sandboxes, with near-native performance on Apple Silicon. Computer provides a PyAutoGUI-compatible interface that can be plugged into any AI agent system (OpenAI Agents SDK , Langchain, CrewAI, AutoGen, etc.). Why Computer? As CUA AI agents become more capable, they need secure environments to operate in. Computer solves this with: • Isolation: Run agents in sandboxes completely separate from your host system. • Reliability: Create reproducible environments for consistent agent behaviour. • Safety: Protect your sensitive data and system resources. • Control: Easily monitor and terminate agent workflows when needed. How it works: Computer uses Lume Virtualization framework under the hood to create and manage virtual environments, providing a simple Python interface: from computer import Computer computer = Computer(os="macos", display="1024x768", memory="8GB", cpu="4") try: await computer.run() # Take screenshots screenshot = await computer.interface.screenshot() # Control mouse and keyboard await computer.interface.move_cursor(100, 100) await computer.interface.left_click() await computer.interface.type("Hello, World!") # Access clipboard await computer.interface.set_clipboard("Test clipboard") content = await computer.interface.copy_to_clipboard() finally: await computer.stop() Features: • Full OS interaction: Control mouse, keyboard, screen, clipboard, and file system • Accessibility tree: Access UI elements programmatically • File sharing: Share directories between host and sandbox • Shell access: Run commands directly in the sandbox • Resource control: Configure memory, CPU, and display resolution Installation: pip install cua-computer GitHub repo: https://github.com/trycua/computer Discord for feedback: https://discord.com/invite/mVnXXpdE85 We're excited to see you building the next Manus general agents with Computer! We'd love to hear your thoughts, feedback, and any questions you might have. What use cases do you see for AI agents running in sandboxes? How do you see Computer being useful in your workflow?
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
Launched alongside, March 2025
the whole month →
Mimic Human Research & Save Findings in AI Knowledge Base
AI · 2025 · sider.ai
