The AI Coding wave is reshaping the paradigm of mobile development, but HarmonyOS faces a more complex situation than iOS and Android. One fundamental contradiction is that the capability boundary of large language models is largely determined by their training corpus. Android has decades of Java and Kotlin accumulation, while iOS has a massive Swift and Objective-C corpus. HarmonyOS’s ArkTS language has only been around for a relatively short time, and publicly available code corpus remains limited.
The challenge goes beyond language. Huawei’s hardware device forms are more diverse, with bar phones, dual-fold devices, and tri-fold devices far exceeding the screen adaptation scenarios of traditional mobile development. The same business code needs differentiated adaptation across different device forms. In addition, HarmonyOS application projects are generally large, and traditional IDEs are designed for human interaction rather than AI workflows. Together, these issues form the real threshold for HarmonyOS AI coding.
H
uawei’s approach is to use engineering tools to compensate for gaps in model capability. DevEco Code is positioned as an out-of-the-box HarmonyOS Coding Agent, covering the full process from project creation, requirements design, and code generation to functional validation and integration testing. DevEco CLI breaks the integrated capabilities of traditional IDEs into an atomic toolset, including command-line capabilities for creating projects, syntax checking, compiling and building, acquiring devices, pushing packages and running, knowledge retrieval, and