skills/jcoger/skills/mobile-scaffold/Gen Agent Trust Hub

mobile-scaffold

Pass

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to perform various build, lint, and test commands as part of the scaffolding and audit gates. Evidence includes the use of 'npx expo export --platform ios' to verify production bundles, 'tsc --noEmit' for quality tooling, and 'eas env:push' for environment management. These commands are well-integrated into the intended developer workflow for mobile application development.
  • [INDIRECT_PROMPT_INJECTION]: In RETROFIT mode, the agent audits existing project files, creating a data ingestion surface where external code (e.g., source code and project configurations) could theoretically attempt to influence the agent's behavior. However, the audit gates rely on technical evidence and binary outcomes (such as command exit codes and specific file fields), which significantly mitigates the risk of natural language instruction override.
  • [SAFE]: The skill actively promotes security and reliability best practices. It enforces toolchain pinning to ensure supply chain integrity, provides warnings against inlining sensitive environment variables into client-side bundles, and scaffolds compliance stubs for platform-specific requirements like third-party AI data-sharing consent.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 9, 2026, 02:06 AM