early-access-email-sequence

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

• Third-party content exposure detected (high risk: 0.85). Outsider free text is ingested via runtime web research: the skill performs WebSearch/WebFetch of public pages (e.g., company site content and search-result surfaced LinkedIn/profile/activity snippets) and then uses that extracted prose to generate the [PERSONALIZE] sections for Emails 2 and 4, which are fed into the agent’s LLM context.