Release v1.3.0 · web-infra-dev/midscene

What's Changed

feat(core): make delayAfterRunner configurable as agent parameter by @yuyutaotao in #1820
feat(web-integration): add timeout parameter to connectNewTabWithUrl and connectCurrentTab methods by @quanru in #1838
feat: prioritize inferring as app name when uncertain in Launch action by @frank-mupt in #1843
feat(cli): set defaultViewportScale to 0 by @EAGzzyCSL in #1841
feat: package name comparison logic by @frank-mupt in #1844
feat(shared): added mergeAndNormalizeAppNameMapping helper function for normalizing and merging app name mappings by @frank-mupt in #1847
feat(computer): add @midscene/computer package for PC desktop automation by @quanru in #1734
feat(all): introduce useDeviceTimestamp args by @frank-mupt in #1857
feat(android): High-performance scrcpy screenshot support by @quanru in #1842

fix(core): context in test cases by @EAGzzyCSL in #1824
fix(core): replay scripts with error by @yuyutaotao in #1827
fix(core): fix Launch and runAdbShell parameter types by @yuyutaotao in #1825
fix(android): improve text input handling for Unicode and special characters by @quanru in #1833
fix(report): remove conflicting height property to prevent timeline layout expansion by @quanru in #1836
fix(ios): remove environment check to support remote WDA by @quanru in #1834
fix(core): forbid 'complete-task' action-type and resolve mixed action/complete-task responses by @yuyutaotao in #1837
fix(android): enhance screenshot validation for problematic devices by @quanru in #1816
fix(shared): include svg tag index in xpath cache for multiple icons by @quanru in #1828
fix(puppeteer): prevent screenshot hang on inactive tabs by @quanru in #1862
fix(ci): improve AI test stability and CI workflow reliability by @quanru in #1855
fix(tests): fix AI test timeout and planning input reliability by @quanru in #1868

refactor(core): change LLM planning prompt output to XML format by @yuyutaotao in #1814
chore(docs): agent.runYaml only parse tasks field by @EAGzzyCSL in #1839
refactor(core): expose sleep in action space by @yuyutaotao in #1840
feat(playwright): append browser name to test titles in multi-project reports by @Copilot in #1835
chore(core): add auto-glm license by @EAGzzyCSL in #1846
chore(ci): use GitHub variables for model configuration by @quanru in #1850
chore(ci): upgrade Node.js version from 18 to 22 (LTS) by @quanru in #1854
chore(docs): add AutoGLM insight model configuration guidance by @EAGzzyCSL in #1856
Gate sub-goals feature behind deepThink flag by @yuyutaotao in #1860
chore(core): log usage when failed to parse content from model by @yuyutaotao in #1859
refactor(core): update strategy of aiAct by @yuyutaotao in #1853

Full Changelog: v1.2.2...v1.3.0