welcome-series

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly requires the agent to "WebFetch the URL" and also fetch related public pages ("/about", "/pricing", "/shop", and product pages), so the agent will ingest and act on arbitrary public website content that can influence its email-writing decisions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs the agent to "WebFetch" the supplied brand URL (e.g., https://brand.com) and its /about, /pricing or /shop and product pages at runtime and uses that fetched content to directly control the generated prompts/output, so these external URLs are runtime dependencies that influence the agent's instructions.