Before execute_e2b_code for attribution, drill-down, or statistical tests, read this SKILL.md then drilldown.md and/or statistical/SKILL.md plus needed references. Does not fetch data.
Scanned 9/8/2026
Install to Claude Code
npx -y skills add infometa/workbuddyskills --skill databrain-analysis --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Databrain Analysis?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/infometa-databrain-analysis)More formats (shields.io, HTML) on the badges page.
---
name: databrain-analysis
display_name_en: DataBrain Sandbox Analysis
display_name_zh: DataBrain 沙箱分析
description: >-
Before execute_e2b_code for attribution, drill-down, or statistical tests,
read this SKILL.md then drilldown.md and/or statistical/SKILL.md plus
needed references. Does not fetch data.
when_to_use: >-
User asks for drivers/归因/下钻/contribution, formal tests (t-test, ANOVA,
regression), or sandbox Python analysis on data already fetched. Not for
final Markdown layout (use databrain-summarize).
---
# DataBrain Analysis (sandbox)
在 **E2B sandbox** 里做归因下钻或统计推断。**不拉数**;先经 `databrain-dashboard-service` / `databrain-intelligence` / opinion skills 取数,再分析。
## 闸门(分析代码前)
用 `read_file` 读引用文件,**再** `open_e2b_code_sandbox` / `upload_to_e2b_code_sandbox` / `execute_e2b_code`。凭记忆写分析代码不算完成。
| 任务类型 | 何时 | 必读 |
|----------|------|------|
| **归因 / 下钻 / 驱动因素** | 贡献度、谁导致变化、多维拆解 | 本文件 + [`drilldown.md`](drilldown.md) |
| **统计检验 / 建模** | A/B、显著性、回归、时序 | 本文件 + [`statistical/SKILL.md`](statistical/SKILL.md) + 按需 `statistical/references/<name>.md` |
| **两者都要** | 先归因定位再对子集做检验 | 先 drilldown,再 statistical |
**未完成闸门**:未读对应 md 已 `execute_e2b_code` 长段分析脚本。
## 数据进 sandbox
1. **优先**:取数工具返回的 **CSV / 本地路径**(`run_skill_script` 写入 `/outputs/...` 或工具说明中的路径)→ `upload_to_e2b_code_sandbox(local_path=...)` → sandbox 内 `pd.read_csv(...)`。
2. **`data_id`**:若工具结果含 `data_id` 且附完整 CSV 路径,上传该文件;勿编造文件名。
3. **`/large_tool_results/` spill**:当工具返回 `Tool result too large` 且路径为 `/large_tool_results/toolu_xxx` 时,**先**按 `soul.md` §7 尝试 **收窄重查**;若分析任务需要保留宽表/全量行且收窄会丢维度,则 **`upload_to_e2b_code_sandbox(local_path=/large_tool_results/...)`**(或 `execute_e2b_code` 引用该路径并由宿主自动 staging)后在 sandbox 内解析。**不要**用 `read_file` 整文件塞进上下文;**不要**在无 `databrain-analysis` 闸门时为简单问答默认走 sandbox。
4. **包安装**:分析前如需 `pandas scipy statsmodels`,先 `execute_e2b_code` `language=bash`:`pip install pandas scipy statsmodels`(勿用 `!pip`)。
## 执行纪律(react)
- `execute_e2b_code`:**短脚本** — 安装依赖 → 读表 → `print()` 表与结论;**禁止** `plt.savefig` / 出图(图表由 **`databrain-chart-render`** 的 `scripts/build_chart.py`:`<dbd>` / echarts_option)。
- **stdout 即证据**:关键统计量、归因表、`ATTRIBUTION_TABLE` 风格列须 `print`;终稿由 **`databrain-summarize`** 合成(含 [`report.md`](../databrain-summarize/report.md) 若 §0.2)。
- 同一 query 内 sandbox **复用**;分析结束可 `close_e2b_code_sandbox`(可选)。
- 工具连续失败 ≥3 次:停止分析,向用户说明原因(`soul.md`)。
## 引用文件
| 文件 | 内容 |
|------|------|
| [`drilldown.md`](drilldown.md) | cumulative_joint、30% 规则、directional/L1、归因结论 |
| [`statistical/SKILL.md`](statistical/SKILL.md) | 选型表、workflow |
| [`statistical/references/`](statistical/references/) | `data_cleaning`, `stat-t-test`, `experimental_design`, … |
| [`../databrain-summarize/report.md`](../databrain-summarize/report.md) | 终稿如何呈现检验/归因(summarize 阶段) |
无 `run_skill_script`(分析逻辑在 sandbox 内手写 Python,可参考 references 中的代码模式)。
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!