engineer-agentic-ai
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes natural-language requests regarding AI behavior (ingestion point) to facilitate modifications of instruction files, skills, and tools (capability inventory). This surface is managed by an engineering framework that requires explicit separation of triggers and actions (boundary markers) and a lifecycle that mandates human review before changes are promoted to production (sanitization).
- [EXTERNAL_DOWNLOADS]: The skill provides references to technical documentation and specifications from trusted sources such as OpenAI, Anthropic, and established industry projects. These resources are for informational guidance and do not facilitate untrusted code execution.
- [COMMAND_EXECUTION]: The reference materials describe the use of local diagnostic CLI tools including
codex,plutil, andjqfor verifying environment state and analyzing rollout logs. These commands are documented for standard engineering diagnostics and troubleshooting within the developer's environment.
Audit Metadata