Allow Cursor to inspect content of dataframes

Ideas Feature Requests
1

Feature request for product/service

AI Models

Describe the request

I frequently run into issues where cursor makes a mistake because it’s not opening and looking at the DataFrames (stored as .pkl) that I’m working with. Would be cool if cursor was able to inspect dataframes in the background and then adjust its code based on the things it finds.

Example: in a pandas dataframe, I have a column of genes some of which are duplicated as gene1-1, gene1-2, … and I would like cursor to join on gene names with another dataframe. without knowledge of the duplication format it will make an erroneous join. inspecting the df beforehand could circumvent that!

2

hi @tdsone and thank you for the feature request.

DataFrames are a binary pickle data format, as you reported with extension .pkl. and in that binary format not readable by AI.

Is there any VScode extension that supports pickle files?
AI could use the files if they are converted from binary to text format.

3

But there is a scenario where the AI has it’s own python interpreter that loads the notebook, prints to stdout and reads that, no?

4

If you instruct AI with the notebook to read it thats a different case, though that would not be in background but in the AI thread in foreground.