native-data-fetching

Pass

Audited by Gen Agent Trust Hub on Sep 6, 2026

Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • Network Operations: The skill implements patterns for making HTTP requests to external services using the fetch API.
  • Evidence: Found in SKILL.md (e.g., fetch('https://api.example.com/users')) and references/expo-router-loaders.md.
  • Context: These operations are central to the skill's purpose. The instructions recommend using non-prefixed environment variables for server-side secrets and SecureStore for client-side token storage to mitigate data exposure risks.
  • Indirect Prompt Injection Surface: The skill provides a framework for ingesting data from external APIs into the agent's context, which could potentially contain untrusted instructions.
  • Ingestion points: Data is fetched via fetch calls and Expo Router loaders in SKILL.md and references/expo-router-loaders.md.
  • Boundary markers: The skill demonstrates typical JSON parsing but does not explicitly show specialized prompt delimiters for the fetched content.
  • Capability inventory: The skill utilizes network capabilities (fetch) and storage access (SecureStore).
  • Sanitization: The documentation in references/expo-router-loaders.md explicitly advises developers to "Validate and sanitize user input... before using in database queries or API calls," which is a proactive security measure.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 6, 2026, 11:17 AM
Security Audit — agent-trust-hub — native-data-fetching