visionos
Pass
Audited by Gen Agent Trust Hub on Feb 28, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill files consist of legitimate documentation and reference guides for visionOS development with no evidence of malicious code or instructions.
- [SAFE]: No signs of obfuscation, hardcoded credentials, or unauthorized network activity were identified in the metadata or body.
- [PROMPT_INJECTION]: The research workflow introduces an indirect prompt injection surface through the retrieval of external content. Ingestion points: Data is gathered from web sources and documentation via the research-expert and apple-docs search tools. Boundary markers: No specific delimiters are specified in the workflow to isolate external retrieved content from system instructions. Capability inventory: The skill focuses on architectural analysis and code implementation without specifying direct system-level execution or data exfiltration capabilities. Sanitization: The workflow does not explicitly mention a process for validating or filtering content retrieved from external sources.
Audit Metadata