Voted for "lean toward consequentialism". As someone once put, I consider the “fundamental” rules to be consequentialist¹, but some of the approximations I use because the fundamental rules are infeasible to calculate from scratch every time resemble deontology or virtue ethics, much like QFT and GR are time-reversal symmetric but thermodynamics isn't. Also, ethical injunctions (i.e. fudge factors in my prior probability that certain behaviours will harm someone to compensate for cognitive biases) and TDT-like game-/decision-theoretical considerations make some of my choices resemble deontology, and a term in my utility function for how awesome I am make some of my choices resemble virtue ethics.

1. I assume that, despite the name, people here don't take consequentialism to imply strictly CDT. I still think that in the True Prisoner's Dilemma against a paperclip maximizer known to use the same decision algorithms as ourselves it's immoral to defect.