Quantum information engines can advantageously convert information into work, usually through a feedback mechanism operated by a Maxwell demon. In principle, this opens prospects of autonomous machines that challenge the classical second law of thermodynamics. Yet prior research, which typically tests a given concept in isolation, usually makes strong assumptions on turning on/off internal interactions or the connection to the baths, which has fueled skepticism. Meanwhile, several recent experiments describe an autonomous spintronic engine that purportedly combines several concepts using inherent electronic interactions to harvest spin fluctuations and quantum correlations. So far, no full-fledged theory has described it. In this article, we address these several interdisciplinary shortcomings by presenting a theory for a solid-state, fully electronic spintronic quantum information engine that converts the energy of quantum coherence into useful electrical work thanks to quantum measurements. Our simple two-stroke engine operates on two correlated atomic, single-spin quantum dots (QDs) that are connected in series with two ferromagnetic electrodes. The ultrafast measurement stroke destroys the quantum coherence stored within the QDs and projects the system into a local high-energy separated state. The resulting energy is then released into the leads as electrical current when the partial thermalizing stroke restores quantum coherence. Using a master equation approach, we show that a robust steady-state current can flow between the electrodes despite the absence of a tunneling path and at thermal equilibrium. Our model proves the feasibility of harvesting quantum fluctuations by the measurement back-action through the on-site Coulomb interaction, and can reproduce experimental output power levels if bosons mediate this transport in the presence of an additional nonequilibrium resource. This can for instance be accomplished by the spin bias that is inherently generated by the ferromagnetic electrodes. Our work raises the prospect of measuring the “spintronic temperature” of the leads using magnetization dynamics experiments, while the model's underpinnings of built-in, materials-inherent measurements of quantum information can in principle be adapted to describe fundamental quantum processes in chemistry and biology.