github.devotd.
NewAI agents

e2b-dev/open-computer-use — Open Source LLM Desktop Sandbox

e2b-dev/open-computer-use drives an E2B Desktop Sandbox with an open source LLM, handling keyboard, mouse and shell commands while streaming the screen in real time.

Repositorye2b-dev/open-computer-use ↗

e2b-dev/open-computer-use drives an E2B Desktop Sandbox with an open source LLM, handling keyboard, mouse and shell commands while streaming the screen in real time.

GitHub description: AI computer use powered by open source LLMs and E2B Desktop Sandbox

stars
★ 2.4K
Forks
272
Language
Python
License
Apache-2.0
Read original source ↗View on GitHub ↗

e2b-dev/open-computer-use is a computer use project that controls an E2B Desktop Sandbox with an open source language model (LLM). The model operates a cloud Linux computer in an isolated sandbox environment on the user's behalf and shows its screen to the user in real time. The structure is one where AI performs the work a person would do with a mouse and keyboard.

The README describes this repository as a project that controls a secure cloud Linux computer based on the E2B Desktop Sandbox with an open source LLM. The control means are the keyboard and mouse together with shell commands, and the sandbox screen is streamed to the client computer in real time. The user can pause the agent at any time and give it new instructions. The default environment is Ubuntu, but the project is designed to work on other operating systems as well. The implementation language is Python, the license is Apache-2.0, and the GitHub star count is 2,377.

The verifiable scope ends at the README feature description and the repository metrics. There are no performance figures or comparison data on the errors and latency that arise while recognizing and operating the screen. Support for other operating systems is presented as a design target, not a verified capability. The only remaining evaluation score is Momentum 3.1 out of 10. The repository was created on 31 October 2024, and no release or change history is recorded, so the improvement history cannot be traced. Performance and maintenance status cannot be judged from the current material alone.