From a tidy table to a first plot

1 · Learn the move · RTO. Role · Task · Output

The single most important word in a figure request is the output: a script, never an image. A figure the model draws in chat is a dead end - you cannot regenerate it when the data updates, cannot show a reviewer how it was made, cannot fix the font without asking again. Role, task, output: the role is a plotting-code author working from your tidy table's schema; the task is one named comparison, not "visualize my data"; the output is a runnable script that reads the file by path and writes the figure to disk. You paste the schema and five rows, not the dataset. The model never sees your data and never needs to - it writes the machinery, your machine does the plotting.

Role: you write plotting scripts; you never draw figures yourself. Task: from a tidy table with this schema - [columns, dtypes, units, factor levels] - plot [the specific comparison]. Here are 5 sample rows: [rows]. Output: a complete Python script using matplotlib that reads the CSV by path, builds the figure, labels axes with units from the schema, and saves to [file.pdf]. No hard-coded data values, no interactive display, no styling I did not ask for.