30-thiet-ke-master

Pass

Audited by Gen Agent Trust Hub on Jun 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill is well-structured and follows secure patterns for asset generation and brand identity management. It uses legitimate API endpoints (api.openai.com) and local project files for its intended purpose.
  • [COMMAND_EXECUTION]: The skill documentation includes logic for detecting local execution environments via environment variables like $OD_BIN. This is used to interface with authorized internal design infrastructure.
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection due to its automated ingestion of external project data.
  • Ingestion points: Reads from project files including prd.md, brand-guideline.md, and campaign briefs as defined in references/brand-identity-source.md.
  • Boundary markers: Absent; the skill does not explicitly wrap ingested data in delimiters within the generated prompts.
  • Capability inventory: The skill can perform external API requests to OpenAI's image generation endpoint and write generated artifacts to the docs/design/ directory.
  • Sanitization: The skill performs structured extraction of brand metadata (hex codes, typography) which limits the attack surface, but it interpolates descriptive text into prompts without explicit sanitization.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 26, 2026, 10:03 AM
Security Audit — agent-trust-hub — 30-thiet-ke-master