1.bytedance / UI-TARS-desktopThe Open-Source Multimodal AI Agent Stack: Connecting Cutting-Edge AI Models and Agent Infraagentvlmvisioncomputer-usemcpmcp-server+8 moreTypeScriptAPACHE-2.040K total4K forks+124 starsfrom 39,386 · score 4.83
2.web-infra-dev / midsceneSee interfaces like a human. E2E testing in natural language.testingbrowser-usecomputer-usegpt-operatorjavascriptphone-use+2 moreTypeScriptMIT15K total1.2K forks+71 starsfrom 15,388 · score 4.28