The process "Extract clusters time series" does not accept two inputs (= it is not available from the Process2 tab). I'd recommend you always generate your script calls to bst_process with the menu "Generate .m script" of the Pipeline editor window.
I also recommend you copy the clusters to all the folders you need with db_set_clusters.m and then refer to them only with their labels in your scripts. See this new features described here:
https://neuroimage.usc.edu/brainstorm/Tutorials/ChannelClusters#Scripting
You script calls should look like this:
sFiles = bst_process('CallProcess', 'process_extract_cluster', sFiles, [], ...
'timewindow', [], ...
'clusters', {'A', 'B'}, ...
'concatenate', 0, ...
'save', 1);
Is there a way to run this same script instead on the subject-averaged files (10 files / 12 files) to get the mean clusters this way
Yes, you can call this process on all the subject-level averages.
To get one matrix file in output for each average (e.g. for computing statistics): you can call the process only once with sFiles including all the files from both groups and the option concatenate=0.
To get one matrix file for each condition, with all the individual subjects concatenated: you need to call the process twice, once for each group, with concatenate=1.
plus the standard deviation?
Is this really the standard deviation across electrodes, that you are interested in? Or across subjects?