qwen-image-pro

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.75). These URLs all point to a single third‑party service (inference.sh) that instructs users to pipe and execute a remote shell installer (curl … | sh) and download platform binaries from its own dist subdomain — a common malware distribution pattern even though HTTPS and checksum files are provided, so it is moderately to highly suspicious unless you fully trust and have independently verified the publisher.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.70). The skill accepts arbitrary external image URIs via the "reference_images" input (see the "Image Editing (Style Transfer)" example and the Input Options table), so it can fetch untrusted public images from third-party URLs which the model will read/interpret and which can materially influence generation behavior.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 1.00). The Quick Start instructs users to run "curl -fsSL https://cli.inference.sh | sh" which fetches and executes remote code from https://cli.inference.sh (and pulls binaries from dist.inference.sh) as a required install step for running the skill, so this external URL can execute remote code at runtime.