Back to directory

HarmonyOS AI Skill

A single markdown knowledge source covering HarmonyOS NEXT app development in ArkTS and ArkUI that automatically generates configuration files for more than eleven AI coding tools, so a general-purpose model stops answering HarmonyOS questions with React or Android conventions.

HarmonyOS AI Skill

HarmonyOS AI Skill packages roughly 4,600 lines of practical HarmonyOS knowledge, distilled from Huawei's official documentation, best practices and API references, into a single markdown source covering strict ArkTS syntax, the Stage application model, more than 60 Kits, and topics from UI state management through persistence. The project argues that general-purpose models rarely see enough ArkTS or HarmonyOS material during training to answer correctly, for example suggesting a React-style useState hook in place of HarmonyOS's own state decorators.

From that one source file, the project generates matching configuration for more than eleven AI coding tools, including Claude Code, Cursor, GitHub Copilot, Codex, Gemini CLI and Windsurf, so installing it once makes each tool answer with HarmonyOS-specific APIs and conventions instead of generic web or mobile ones. Installation for Claude Code is cloning the repository and symlinking its HarmonyOS development folder into the local skills directory; other tools follow a separate installation guide.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.