Open-Dots: An Open-Source, Self-Hosted Workspace for AI Agents
Anil-matcha/open-dots has gained traction on GitHub with over 460 new stars in a week. The project provides an open-source, self-hosted Python workspace for building and running AI agents equipped with chat, approval workflows, and optional desktop control capabilities. As interest in AI agents that can directly interact with computer interfaces grows, open-dots offers a transparent, privacy-focused alternative to proprietary systems like Manus or OpenAI Operator. It enables developers to maintain control over their data, API connectors, and execution environment without vendor lock-in. Distributed under the MIT License, open-dots features built-in chat UI, connectors, workflow approval mechanisms, and optional computer use functionality. The maintainers emphasize that the repository is currently in an early prototype stage.
## BACKGROUND
Computer-use AI agents are autonomous tools capable of reading screen outputs, executing mouse clicks, typing keystrokes, and running terminal commands to perform complex software tasks. Proprietary implementations, such as Anthropic's Claude computer use feature or OpenAI's Operator, automate desktop workflows via cloud models, prompting developers to seek self-hosted open-source alternatives for local execution.