query-tableau-data

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

• Third-party content exposure detected (high risk: 0.75). At runtime the skill calls Tableau’s Metadata API (GraphQL) and VDS endpoints, which return outsider-authored free text such as datasource/workbook/view names, descriptions, and field captions/formulas; this text is then placed into the agent’s LLM context via printed summaries (e.g., print(f"{scope.datasource_count} ..."), print([(v.name, v.total_view_count) ...]), print(f" {p.name}"), and print(result.rows[:5])).