#711 · Primary category： AI Agents & Automation
browserwing
BrowserWing converts browser actions into MCP commands for efficient AI agent control.
Project last updated：08/08/26
GitHub Stars
1.4K
Forks
128
Contributors
5
License
MIT
Why we included this project
Teams building AI agents often burn tokens on screenshot-heavy loops just to get a browser to do something simple. BrowserWing sidesteps that by exposing browser actions as callable commands over the Model Context Protocol, so an agent can drive the page and pull data without narrating each step. It ships with 78 ready-made scripts that return structured JSON from sites like GitHub trending, Hacker News, Bilibili, and Reddit, plus a visual recorder that turns recorded actions into MCP commands or a Skills file. A CLI mode lets agents and shell pipelines call those scripts directly and pipe the JSON out, which is handy for headless, scriptable scraping. If your work involves web scraping or feeding live page data back into an agent loop, this makes those interactions faster and more predictable.
Articles for this project
No articles for this project yet.
To suggest a topic or contribute an article, contact us.
Related projects in this category
openclaw
Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞
hermes-agent
The agent that grows with you
n8n
Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.
deepseek-harness
DeepSeek Harness: Everything is a Plugin.
AutoGPT
AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on what matters.