v0.27.0
What's Changed
New Features 🎉
- feat: update yaml cache for action space by @yuyutaotao in #1063
- feat(web-integration): let aiAssert support domIncluded and screenshotIncluded opt by @quanru in #1086
- feat(report): the report supports hiding the marker overlay. by @quanru in #1083
- feat(web-integration): add optional name field to aiAssert for JSON output and update related documentation by @quanru in #1089
- feat(core): enable override modle config when create agent by @EAGzzyCSL in #1002
- feat(web): aiWaitFor support image prompting by @EAGzzyCSL in #1093
- feat(report): display multi model info in report by @EAGzzyCSL in #1103
- feat(report): display intent with model name by @EAGzzyCSL in #1104
Bug Fixes 🐞
- fix(web-intergration): clear dump annotation description after processing to optimize memory usage by @quanru in #1075
- fix(chrome-extension): delete useless host_permissions by @quanru in #1078
- fix(web-integration): keyboard press schema by @quanru in #1082
- fix aiWaitFor throw error unexceptly by adding doNotThrowError option by @quanru in #1085
- fix(core): action player by @yuyutaotao in #1097
- fix(core): ui-tars planning by @yuyutaotao in #1101
Document 📖
Other Changes
- refactor(core): use the same implementation for aiWaitFor as in aiBoolean by @EAGzzyCSL in #1071
- refactor(core): dump model_name in group level by @EAGzzyCSL in #1077
- use zod to define params for action space by @yuyutaotao in #1080
- refactor(web-integration): update aiInput and aiKeyboardPress handling to support new format by @quanru in #1087
- chore(shred): rename resizeImgBuffer to resizeAndConvertImgBuffer by @EAGzzyCSL in #1092
- refactor(core): put locate in action param by @yuyutaotao in #1088
- chore(core): recover describe function by @EAGzzyCSL in #1096
- refactor(shared): refactor implements of vlLocateMode and uiTarsModel… by @EAGzzyCSL in #1102
Full Changelog: v0.26.6...v0.27.0