Abstract We compare several common subgrid implementations of active galactic nucleus (AGN) feedback, focusing on the effects of different triggering mechanisms and the differences between thermal and kinetic feedback. Our main result is that pure thermal feedback that is centrally injected behaves differently from feedback with even a small kinetic component. Specifically, pure thermal feedback results in excessive condensation and smothering of the AGN by cold gas because the feedback energy does not propagate to large enough radii. We do not see large differences between implementations of different triggering mechanisms, as long as the spatial resolution is sufficiently high, probably because all of the implementations tested here trigger strong AGN feedback under similar conditions. In order to assess the role of resolution, we vary the size of the “accretion zone” in which properties are measured to determine the AGN accretion rate and resulting feedback power. We find that a larger accretion zone results in steadier jets but can also allow too much cold gas condensation in simulations with a Bondi-like triggering algorithm. We also vary the opening angle of jet precession and find that a larger precession angle causes more of the jet energy to thermalize closer to the AGN, thereby producing results similar to pure thermal feedback. Our simulations confirm that AGNs can regulate the thermal state of cool-core galaxy clusters and maintain the core in a state that is marginally susceptible to thermal instability followed by precipitation.