English
Basic Excel Data Quality Check
Perform a read-only spreadsheet data-quality audit using generic public-safe rules.
Safety Boundary
- Do not include finance or investment advice.
- Do not infer business meaning without source evidence.
- Do not clean, standardize, fill, deduplicate, overwrite, or generate a modified spreadsheet copy.
- Cite sheet names, column names, row ranges, and field names where possible.
Routing Boundary
- Use this Skill for requests to check, audit, or list missing values, duplicates, types, formats, and anomalies without changing data.
- Use
spreadsheet-data-cleaningwhen the user asks to clean, normalize, transform, repair, fill, deduplicate, or create a cleaned output. - If a request starts as an audit but asks for changes, do not apply them; hand the modification task to
spreadsheet-data-cleaning. - When spreadsheet tools are unavailable, report what cannot be inspected and do not claim that a data-quality check was completed.
Workflow
- Identify workbook structure: sheets, tables, headers, dimensions, and data types.
- Check missing values, duplicate rows, inconsistent formats, impossible values, and mixed units.
- Check whether key fields and source fields are present.
- Summarize data quality risks and manual confirmation items.
- Suggest safe next steps without applying changes or generating a modified file.
Output Format
- Workbook overview
- Sheet and field inventory
- Data quality findings
- Evidence references
- Suggested next steps not applied
- Manual confirmation items
简体中文
Excel 数据质量基础检查
使用通用、公开安全的规则执行只读表格数据质量审计。
安全边界
- 不包含财务或投资建议。
- 没有来源证据时不推断业务含义。
- 不清洗、标准化、填补、去重、覆盖或生成修改后的表格副本。
- 尽量引用工作表名称、列名、行范围和字段名。
路由边界
- 用户要求在不改变数据的前提下检查、审计或列出缺失、重复、类型、格式和异常时使用本 Skill。
- 用户要求清洗、规范化、转换、修复、填补、去重或生成清洗结果时使用
spreadsheet-data-cleaning。 - 如果请求先要求审计又要求修改,不在本 Skill 中执行修改;把修改任务转交
spreadsheet-data-cleaning。 - 表格工具不可用时报告无法检查的内容,不得声称已完成数据质量检查。
工作流程
- 识别工作簿结构:工作表、表格、表头、维度和数据类型。
- 检查缺失值、重复行、格式不一致、不可能值和单位混用。
- 检查关键字段和来源字段是否存在。
- 总结数据质量风险和人工确认项。
- 建议安全后续步骤,但不应用修改或生成修改后的文件。
输出格式
- 工作簿概览
- 工作表和字段清单
- 数据质量问题
- 证据来源
- 未执行的建议后续步骤
- 人工确认项