fantasy-design

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • [PROMPT_INJECTION]: The skill ingests user-defined tasks and descriptions to generate complex visual parameters, creating a surface for potential indirect prompt injection if untrusted data is processed.
  • Ingestion points: User mission requirements and scene descriptions are processed to determine appropriate visual presets (SKILL.md).
  • Boundary markers: The system uses a structured 10-section output protocol which helps delimit generated content from instructions (COMMON-12-SECTION-PROTOCOL.md).
  • Capability inventory: The skill produces style routes and modules for downstream visual generation agents, which could be subverted via poisoned input.
  • Sanitization: Includes a quality gate (Section 8) and a negative case library (NEGATIVE-CASE-BOOK.md) to prevent common generation failures.
  • [EXTERNAL_DOWNLOADS]: The skill references educational content and tutorials hosted on Douyin (SOURCE-LEDGER.md), which is a well-known platform. These references are used neutrally to provide context for the design methodologies described in the skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 02:31 PM
Security Audit — agent-trust-hub — fantasy-design