mind-weapon

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

• Third-party content exposure detected (high risk: 0.75). The required workflow ingests outsider-authored free text primarily from user-provided inputs and assets—e.g., onboarding attachments fetched and extracted at runtime (guides/06-onboarding-flow.md:98-206) and arbitrary scraped URL text (guides/06-onboarding-flow.md:131-137)—which can then be concatenated into the agent’s prompt/context; while SKILL.md is doc-owned, the runtime path described by the guides includes these outsider sources.