spree-i18n

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

• Third-party content exposure detected (high risk: 0.85). The skill’s “Fetch live docs” section instructs runtime fetching of public web pages (e.g., https://spreecommerce.org/..., https://guides.rubyonrails.org/..., and GitHub README), whose readable text could be ingested into the agent’s LLM context as free-form content.

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

• Potentially malicious external URL detected (high risk: 0.70). The skill explicitly instructs the agent to fetch live documentation at runtime (e.g., https://spreecommerce.org/docs/developer/core-concepts/translations and https://github.com/spree-contrib/spree_i18n) to drive how it writes code and verifies behavior, meaning external content would directly influence the agent's instructions.