Executed Notebook Tools

Configuration File: executed_notebook_tools.json Tool Type: Local Tools Count: 1

This page contains all tools defined in the executed_notebook_tools.json configuration file.

Available Tools

read_executed_notebook (Type: ExecutedNotebookTool)

Read a pre-executed Jupyter notebook (.ipynb, often named ‘*_executed.ipynb’) and return its cell…

read_executed_notebook tool specification

Tool Information:

  • Name: read_executed_notebook

  • Type: ExecutedNotebookTool

  • Description: Read a pre-executed Jupyter notebook (.ipynb, often named ‘*_executed.ipynb’) and return its cells with printed outputs as structured JSON. Use this BEFORE reimplementing an analysis when the data folder ships with an executed notebook — the notebook already ran the authoritative pipeline with the exact package versions, filters, and thresholds that match the reference answers. Reimplementing with pydeseq2/scanpy/etc. produces different numbers. Accepts either a folder (auto-discovers the notebook) or a direct path. Optionally filters cells by a comma-separated search term list or a regex so the agent sees only the cells relevant to its question.

Parameters:

  • data_folder ([‘string’, ‘null’]) (optional) Directory to search (recursively) for a .ipynb file. Prefers *_executed.ipynb over regular .ipynb. Provide either this OR notebook_path.

  • notebook_path ([‘string’, ‘null’]) (optional) Direct path to a specific .ipynb file. Provide either this OR data_folder.

  • search (string) (optional) Optional: comma-separated terms (or regex if regex=true). Cells whose source or output contains any term are returned in ‘matching_cells’. Case-insensitive. Example: ‘upregulated,log2FoldChange’ or ‘padj.*0.05’

  • regex (boolean) (optional) Treat ‘search’ as a Python regex instead of a comma-separated term list (default: false)

  • max_output_chars (integer) (optional) Truncate each cell source/output to this many characters (default 400)

  • include_source (boolean) (optional) Include cell source code in the response (default true). Set false to halve response size when you only need outputs.

Example Usage:

query = {
    "name": "read_executed_notebook",
    "arguments": {
    }
}
result = tu.run(query)