medusa-customers

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

• Third-party content exposure detected (high risk: 0.75). The required workflow includes “Fetch live docs” via runtime web-search and fetching Medusa documentation URLs, which can ingest outsider-authored free text from public web pages into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs the agent at runtime to "Fetch https://docs.medusajs.com/resources/references/customer" to review the ICustomerModuleService interface, meaning external content from that URL would be fetched and used to control the agent's implementation/prompting.