svelte

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime workflow ingests outsider-authored code snippets a user provides to the @sveltejs/mcp CLI tools (e.g., npx @sveltejs/mcp svelte-autofixer "<code_or_path>"), meaning free text authored by a provider user can be read by the LLM during analysis/autofixing.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs running npx to fetch and execute the external CLI package (npx @sveltejs/mcp), which will download and run remote code at runtime and is described as a required tool for the skill.