brand-logos

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s REST API workflow ingests untrusted free-text indirectly by sending the attacker-controlled user query to GET https://api.svgl.app?search=<query> and then fetching the returned SVG markup from GET https://api.svgl.app/svg/<name>.svg (including inline usage), so outsider-authored text can influence which content is read at runtime.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill instructs the shadcn CLI to fetch remote registry data at https://svgl.app/r/{name}.json at runtime to generate/install React components, so remote JSON controls code generation used by the agent.