using-react-native-hifi
Pass
Audited by Gen Agent Trust Hub on Apr 8, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill employs forceful instructional language (e.g., "YOU ABSOLUTELY MUST", "Not negotiable") to ensure the agent invokes relevant skills before providing responses. This steering is designed for workflow discipline and explicitly states that user-provided instructions (like CLAUDE.md or GEMINI.md) retain the highest priority over the skill's directives.
- [COMMAND_EXECUTION]: Contains standard Git utility commands (e.g.,
git rev-parse,git branch --show-current) used for environment detection to ensure the agent correctly handles worktrees and detached HEAD states. - [SAFE]: Provides detailed tool mapping references that translate Claude Code-specific tools to their equivalents in Codex, Copilot CLI, and Gemini CLI, ensuring the agent uses the correct platform-native capabilities.
- [SAFE]: Includes guidance on subagent dispatch and prompt framing using XML tags, which follows established prompt engineering best practices for reliable instruction adherence.
Audit Metadata