hmos-arkui-mvvm-pattern

Pass

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: SAFECREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: Hardcoded keyPassword and storePassword fields are present in the build-profile.json5 configuration files within the test-cases/MVVMTestCaseV1/ and test-cases/MVVMTestCaseV2/ directories. These values are part of the standard HarmonyOS build configuration for local debug signing and do not represent production credentials.
  • [PROMPT_INJECTION]: The skill acts as an architecture assistant that reviews and refactors ArkUI code provided by the user. This ingestion of untrusted external content represents a surface for indirect prompt injection if the user-provided code contains instructions designed to bypass agent constraints.
  • Ingestion points: User-provided ArkTS and ArkUI code snippets intended for refactoring or architectural review.
  • Boundary markers: Not explicitly implemented in the skill instructions.
  • Capability inventory: The skill is designed to modify .ets files and utilize development tools for project building and static analysis.
  • Sanitization: No specific filtering or validation of user-provided logic is mentioned in the skill implementation.
  • [SAFE]: The skill references external image assets from picsum.photos for use as placeholders in the Article Detail page templates. This is a well-known service for development and UI prototyping.
  • [SAFE]: The use of HarmonyOS-specific decorators (e.g., @Observed, @Trace, @Param) and system APIs (e.g., AppStorageV2, PersistenceV2) for state management and resource handling adheres to official documentation and the primary purpose of the skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 22, 2026, 10:17 AM
Security Audit — agent-trust-hub — hmos-arkui-mvvm-pattern