Long-duration gamma-ray bursts are associated with the collapse of massive stars, occurring ubiquitously in star-forming environments. These insights are based on studies of the more than 1000 bursts localised over the past 20 years. However, last year one burst, GRB 191019A, was discovered which appeared to challenge this rule. GRB 191019A is a regular long GRB coincident with the nucleus of an apparently passive galaxy at z=0.248. The probability of chance alignment within 0.1" of the galaxy nucleus is minimal (~10^-6), and so the association appears secure. There is no sign of any underlying supernova (SN) to limits >10 times fainter than typically in GRB supernovae. There is also no indication of nuclear SMBH activity in the galaxy (no X-ray detection to deep limits, or emission lines in the optical spectrum), and the source vanishes within a few hours, too quick for the expected evolution of tidal disruption events. This event may therefore represent a genuinely new route to the creation of a long GRB. Here we request deep UV observations to search for any underlying star formation to limits of <0.01 Msol/yr. The detection of even weak star formation would favour the origin in a massive star, in keeping with other GRBs. In this case, it is likely to provide evidence for the direct collapse of a massive star to a black hole. Alternatively, the absence of star formation may suggest a dynamical channel in the dense nuclear regions of the host galaxy. These HST observations are hence crucial in determining the origin of this new, and possibly so-far unidentified signal of stellar death....