gpt-image-prompts

Pass

Audited by Gen Agent Trust Hub on Aug 30, 2026

Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: No security issues were detected. The skill is entirely instructional and serves as a reference for prompt engineering.
  • [SAFE]: The skill links only to official OpenAI documentation for GPT Image models, which is a trusted source.
  • [NO_CODE]: No scripts (Python, JS, etc.) or external package dependencies are present, eliminating potential attack vectors related to remote code execution or supply chain risks.
  • [INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface for user-provided visual ideas, but no capabilities to exploit. 1. Ingestion points: User creative briefs are processed as instructions for generating image prompts. 2. Boundary markers: Present; the skill uses structured markdown templates and labeled sections. 3. Capability inventory: None; tool invocation is disabled and no scripts are included. 4. Sanitization: Not applicable as the output is static text with no execution context.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 30, 2026, 06:31 PM
Security Audit — agent-trust-hub — gpt-image-prompts