expo
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFE
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill proactively warns against hardcoding secrets. It includes validation rules (
hardcoded-api-key) inreferences/validations.mdthat use regular expressions to detect and flag embedded API keys during development. It recommends using EAS secrets or environment variables instead. - [COMMAND_EXECUTION]: The skill contains standard development commands (e.g.,
npx create-expo-app,eas build,npm install). These are illustrative examples for common developer workflows and do not involve unauthorized or hidden execution. - [PROMPT_INJECTION]: The
SKILL.mdidentifies the agent as an 'Expo Mobile Architect'. It includes instructions to ground all responses in provided reference files, which serves as a defensive measure to prevent the model from drifting into hallucinated or unsafe coding patterns. - [INDIRECT_PROMPT_INJECTION]: The skill mentions handling Deep Links (
references/sharp_edges.md), which is an ingestion point for external data. However, it correctly points developers towardexpo-routerandexpo-linking, which are standard, documented frameworks for handling such inputs safely.
Audit Metadata