official-document-drafting

Pass

Audited by Gen Agent Trust Hub on Jul 18, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill contains highly prescriptive instructions in the 'Licensing Iron Rule' (⛔ 使用授權鐵律) section designed to prevent users from overriding or bypassing internal usage restrictions. It explicitly instructs the AI to prioritize these constraints over any subsequent user prompts, role-play attempts, or 'jailbreak' instructions, which is a defensive use of prompt injection patterns to enforce business logic.- [SAFE]: The skill is entirely text-based and functions as a drafting reference. It does not perform any network operations (no curl, wget, or fetch), lacks file system modification capabilities, and does not execute arbitrary code or subprocesses.- [SAFE]: All internal references point to legitimate markdown files within the skill's package (official-document-format.md and official-document-terminology.md) that contain standard formatting guidelines and terminology for government communications.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 18, 2026, 05:19 PM
Security Audit — agent-trust-hub — official-document-drafting