Title: Proposal for a clumsiness-free test of macroscopic realism

Abstract: We propose a test of macrorealism that exploits the contextuality of two-time correlation functions to escape the so-called "clumsiness loophole" that plagues Leggett-Garg inequalities. The non-contextuality of reduced joint probability distributions is proven to be an unequivocal criterion to guarantee that measurements are carried out in the ideally-weak measurement regime of a class of generalized von Neumann measurements. In this regime, testing the so-called "no-signaling in time" condition allows to uncontextually ascertain whether a property of a given system is macrorealistic or non-macrorealistic. The resulting protocol allows for tests of macrorealism in situations where Leggett-Garg inequalities and ideal negative measurement cannot be used at all.