How organs reach their final shape is a central yet unresolved question in developmental biology. Here we investigate whether mechanical cues contribute to this process. We analyze the epidermal cells of the Arabidopsis sepal, focusing on cortical microtubule arrays, which align along maximal tensile stresses and restrict growth in that direction through their indirect impact on the mechanical anisotropy of cell walls. We find a good match between growth and microtubule orientation throughout most of the development of the sepal. However, at the sepal tip, where organ maturation initiates and growth slows down in later stages, microtubules remain in a configuration consistent with fast anisotropic growth, i.e., transverse, and the anisotropy of their arrays even increases. To understand this apparent paradox, we built a continuous mechanical model of a growing sepal. The model demonstrates that differential growth in the sepal can generate transverse tensile stress at the tip. Consistently, microtubules respond to mechanical perturbations and align along maximal tension at the sepal tip. Including this mechanical feedback in our,growth model of the sepal, we predict an impact on sepal shape that is validated experimentally using mutants with either increased or decreased microtubule response to stress. Altogether, this suggests that a mechanical feedback loop, via microtubules acting both as stress sensor and growth regulator, channels the growth and shape of the sepal tip. We propose that this proprioception mechanism is a key step leading to growth arrest in the whole sepal in response to its own growth.