data-export

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime import path reads outsider-authored free text from a user-selected file URL (e.g., JSON/CSV content) via Data(contentsOf: url) / CSVExporter.parse(from: url), which then becomes LLM-readable strings (e.g., parsed CSV rows) and can be used in subsequent agent reasoning.