scan-site

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

• Third-party content exposure detected (high risk: 0.85). The skill fetches an outsider-authored deep-scan report from a public service API at runtime (get-latest-report.js → transform-report.js), and the report’s plain-language fields (e.g., AlertName, Description, Mitigation, LearnMoreLink) are parsed and inserted into the agent’s LLM context via the rendered/plain-language summary.