search-intent-coverage

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 fetching live SERPs and/or opening competitor webpages (via a SERP API and Browser MCP or the agent-browser CLI) and instructs the agent to extract and act on headings/content from those public sites, so untrusted third‑party content is read and used to drive decisions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill requires a SERP API and explicitly opens/fetches arbitrary competitor pages at runtime using the agent-browser command (agent-browser open ""), which pulls remote HTML that is injected into the agent's context and therefore directly controls prompt content.