medusa-storefront

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md "Before writing code" section explicitly instructs the agent to fetch live docs from https://docs.medusajs.com and to web-search GitHub for the Next.js starter, so the agent will ingest public third-party (including user-generated GitHub) content that can directly influence implementation decisions and tool use.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill explicitly includes storefront checkout and payment actions: it references a publishable API key, a "Payment provider session", and SDK methods such as sdk.store.cart.initiatePaymentSession() and sdk.store.cart.complete() for initiating payment sessions and completing orders. Those are specific, purpose-built APIs to process customer payments (i.e., move money) rather than generic tooling. Therefore it grants direct financial execution capability.