data

# 数据诊断 先获取 schema 和必要样本，报告表规模、字段类型、缺失、重复、异常、时间范围和关键分布。区分事实、风险与建议，指出可直接分析的字段及需要用户确认的口径。 ## Tool routing 1. Use `get_schema` first to inspect tables, fields, row counts, and source structure. 2. Use `get_table_detail` when one table needs deeper field-level metadata. 3. Use `profile_data` for data quality, missingness, type, and distribution diagnostics. 4. Use `query_data` for small verification samples or targeted aggregates. 5. Use `generate_chart` only when a compact diagnostic chart materially helps explain the data. ## Implementation reference - Data tool entries: `agent/tools/business/data.py` - Profiling implementation: `Function/Clean/data_profile.py` - Chart implementation: `Function/Charts_generation/chart_generate.py`