gpt-image-2-gen

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

• Third-party content exposure detected (high risk: 0.80). The skill accepts arbitrary public image URLs as reference images (see SKILL.md "Reference images" and scripts/gpt-image-gen.sh --image / references/api-params.md image_urls), which are untrusted third-party content that the agent includes in generation/editing workflows and can materially influence model behavior, so it exposes the agent to potential indirect prompt injection via those external images.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's installation/agent instructions tell an agent to run remote installers (e.g., "npx evolink-gpt-image@latest -y" and git install via https://github.com/EvoLinkAI/gpt-image-2-gen-skill-for-openclaw), which cause the agent to fetch and execute remote code at runtime (installer), so this external URL is a runtime-executed dependency: https://github.com/EvoLinkAI/gpt-image-2-gen-skill-for-openclaw