Skills
Navigate
skills/osidemedia/higgsfield-ai-prompt-skill/higgsfield-gpt-image-2/Gen Agent Trust Hub

higgsfield-gpt-image-2

Pass

Audited by Gen Agent Trust Hub on Oct 26, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: Processes user input and product data, creating an indirect prompt surface.
  • Ingestion points: User input in SKILL.md and product pages in static-ads-workflow.md.
  • Boundary markers: Structured JSON and quoted text serve as delimiters.
  • Capability inventory: Primarily image prompt generation.
  • Sanitization: Recommends extracting substance from data.
  • [EXTERNAL_DOWNLOADS]: References official documentation for model details.
  • Evidence: Documentation links to platform.openai.com in SKILL.md.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 26, 2026, 11:52 AM