human-interface-guidelines
Pass
Audited by Gen Agent Trust Hub on Mar 4, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The script
scripts/fetch_hig_page.pyfetches JSON data fromdeveloper.apple.com. This is a trusted and well-known documentation service, and the script restricts access to official subdomains. - [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection by retrieving and processing external content from the Apple developer portal.
- Ingestion points: Documentation is ingested through the
scripts/fetch_hig_page.pyscript. - Boundary markers: The script outputs Markdown content directly without using delimiters or system instructions to isolate the retrieved text from the agent's logic.
- Capability inventory: The skill includes Python scripts capable of searching local catalogs and performing network requests.
- Sanitization: No sanitization is performed on the retrieved documentation to filter out potential instructions before it is presented to the agent.
Audit Metadata