shadcn-tailwind

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

• Third-party content exposure detected (high risk: 0.80). The skill's Installation & CLI and references/components.md explicitly instruct using commands like "npx shadcn add/view/search" and defines components.json "registries" with external URLs (e.g., https://acme.com/r/{name}.json), meaning the agent will fetch and install components from public community registries and thus ingest untrusted third-party content that can change subsequent actions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill includes runtime CLI operations (e.g., "npx shadcn add @acme/button @internal/auth-system") that will fetch component manifests/templates from configured registries such as https://acme.com/r/{name}.json and https://registry.company.com/{name}, which are fetched at runtime and supply remote templates/code used to generate project files (i.e., remote content that directly controls the tool's behavior).