extension-create

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly tells the agent to "Browse to the relevant sample on GitHub" and "Copy the core logic" in references/chrome-samples-reference.md, which requires fetching and interpreting public, user-generated GitHub content that could materially influence generated code and actions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill's scaffold step explicitly tells the operator to run commands that fetch-and-execute remote packages at runtime (e.g., "npx wxt@latest init " which pulls/executes WXT from npm — related docs at https://wxt.dev/ — and "npm create plasmo ..." which pulls/executes Plasmo — docs at https://docs.plasmo.com/), so required external content is executed during runtime.