taste-imagegen-frontend-web
Pass
Audited by Gen Agent Trust Hub on Jul 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is entirely composed of markdown instructions and configuration intended to guide image generation tasks. It does not include any scripts, dependencies, or commands that could lead to unauthorized access, data exfiltration, or malicious code execution.
- [PROMPT_INJECTION]: While the skill contains strong instructional language (e.g., "HARD OUTPUT RULE", "This rule overrides any model default"), these are used exclusively to enforce formatting constraints on the output (generating one image per section) and do not attempt to bypass AI safety guidelines or extract system information.
- [DATA_EXFILTRATION]: There are no network calls, hardcoded credentials, or instructions to access sensitive local files. The skill operates within the context of generating visual content based on user prompts.
- [REMOTE_CODE_EXECUTION]: No external scripts are downloaded or executed. The skill does not use package managers or dynamic code execution environments.
Audit Metadata