software-ios-ai-engine

Pass

Audited by Gen Agent Trust Hub on Oct 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes scripts/scaffold-composers.sh, a shell utility designed to generate Swift boilerplate for developers. The script uses standard mkdir and printf commands to write static template strings into a directory specified by the user. It does not perform network requests, modify system configurations, or access sensitive user data.
  • [INDIRECT_PROMPT_INJECTION]: The skill outlines a multi-tier architecture for processing untrusted user input and retrieval data. It explicitly addresses injection risks through a 'Tier 0' router that employs regex-based safety boundaries and crisis redirection before any model processing occurs. Furthermore, it mandates a post-processing stage that validates model output against a whitelist of evidence anchors and word counts to prevent hallucination and instruction override.
  • [EXTERNAL_DOWNLOADS]: The data/sources.json and references/ files include numerous links to external resources. These primarily target official Apple Developer documentation, Apple Machine Learning Research, and established technology platforms such as GitHub and arXiv. These references are for documentation purposes and do not involve the automated download or execution of remote code at runtime.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 6, 2026, 02:03 PM