We deliberately do not expose clusters since they are local to the histogrammer and only meant to cross-check results quickly.
If you need more elaborate clustering and tracking I would suggest storing the output in the Corryvreckan format using the CorryvreckanWriter. You can then do all the analysis including one-track/off-track selection there.
We might have to improve a bit how MCParticles are used there, but maybe first have a look and we can then adjust
Dear @simonspa , thanks a lot for the feedback.
Yes, it makes sense that clusters are available after a reconstruction step. I will have a look at Corryvreckan documentation.