菜单
Explore datasets in Jupyter notebooks by inspecting dataframes, summarizing schemas, checking data quality, and proposing EDA cells. Use when asked to analyze data or add exploratory analysis.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Explore datasets in Jupyter notebooks by inspecting dataframes, summarizing schemas, checking data quality, and proposing EDA cells. Use when asked to analyze data or add exploratory analysis.
Diagnose and minimally fix failing notebook cells. Reads the failing cell, inspects the error output, checks dependencies, applies the smallest fix, and reruns only what is needed.
Make Jupyter notebooks easier to read and share. Identifies duplicated cells, improves structure, adds markdown documentation, and prepares notebooks for collaboration.
Explain what a Jupyter notebook does by inspecting structure, summarizing workflow, and reading only key cells. Use when asked to walk through, summarize, or understand a notebook.
Audit Jupyter notebooks for hidden state, stale outputs, execution order issues, and non-determinism. Produces a report with findings, confidence labels, and suggested fixes.
Improve or debug plots in Jupyter notebooks. Finds plotting cells, critiques visualizations, applies minimal fixes, and reruns only the affected cells.
| name | eda-copilot |
| description | Explore datasets in Jupyter notebooks by inspecting dataframes, summarizing schemas, checking data quality, and proposing EDA cells. Use when asked to analyze data or add exploratory analysis. |
| license | MIT |
| compatibility | opencode |
| metadata | {"audience":"data-scientists","workflow":"analysis"} |
I help explore and understand datasets loaded in Jupyter notebooks. I:
If in live mode, list variables:
notebook-tools list-variables --notebook <path> --start-if-missing --pretty
Or search for data-loading patterns:
notebook-tools search-cells --notebook <path> --query "read_csv" --pretty
notebook-tools inspect-dataframe --notebook <path> --variable-name <name> --pretty
This returns shape, columns, dtypes, null counts, and sample rows.
notebook-tools read-cells --notebook <path> --index <N> --summary-mode compact --pretty
Based on the dataframe inspection, propose:
If the user wants EDA cells inserted, use:
notebook-tools insert-cell --notebook <path> --position <N> --cell-type code --content "<eda-code>"
Ask before inserting new cells into the notebook.
Requires notebook-tools CLI. Live inspection requires ipykernel, jupyter_client, and pandas.