The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the belt CLI tool (via the Bash tool) to perform operations such as belt login and belt app run. These are legitimate actions for interacting with the Inference.sh inference service.
[EXTERNAL_DOWNLOADS]: The skill references external resources including the inferencesh Python package and the belt-sh/cli skill. It also links to installation guidelines hosted on the official inference-sh GitHub repository. These resources are consistent with the skill's stated purpose of providing an interface to the Alibaba Qwen model via Inference.sh.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided strings through the prompt parameter. While this is an ingestion point for untrusted data, the skill acts as a pass-through to an image generation API, which typically has its own internal safety filters for content generation. The risk of indirect injection affecting the agent's logic is minimal.

qwen-image-2-pro