qwen-image-2-pro
Pass
Audited by Gen Agent Trust Hub on Mar 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
infshCLI tool (a verified vendor resource) via the Bash tool to execute image generation tasks on the Inference.sh platform. - [EXTERNAL_DOWNLOADS]: The documentation references external images and guidelines hosted on
inference.shandcloud.inference.sh. It also suggests installing related skills usingnpx. - [PROMPT_INJECTION]: The skill has an indirect prompt injection surface.
- Ingestion points: User-provided prompts and negative prompts are passed directly into the tool input in
SKILL.md. - Boundary markers: Input is structured within a JSON object, providing basic separation from the command execution context.
- Capability inventory: The skill can execute shell commands (
Bash) and make network requests via theinfshutility. - Sanitization: No internal sanitization or validation logic is present in the skill's instructions; filtering is assumed to be handled by the backend model.
Audit Metadata