modern-web-guidance

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow is search/retrieve from the modern-web-guidance npm package, which necessarily fetches and ingests guide text from its own guide corpus (i.e., non-user-authored content) into the agent’s LLM context via the retrieve command output.

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

• Potentially malicious external URL detected (high risk: 0.80). The SKILL requires running the npx command "npx -y modern-web-guidance@latest" (i.e., fetching and executing the modern-web-guidance npm package at runtime), which pulls remote code from the npm registry and therefore executes external content that directly controls the skill's behavior.