Abstract Shade trees in agroforestry systems confer ecosystem services, such as enhanced soil fertility from diverse litter inputs, microclimate regulation via shade, and disease mitigation through trophic and abiotic interactions. With this thriving role of agroforestry in sustainable agriculture, particularly for tree crops, systematic and reliable methods to select shade trees for specific agroecosystem outcomes are crucial. Plant functional traits offer a framework to describe, select and manage shade trees. Over the last decade, shade tree leaf functional traits and whole plant traits have been assessed in agroforestry systems. Yet, we lack amalgamated information on (i) what we know about shade tree trait relationships with functions to achieve desired agroecosystem outcomes, (ii) how decades of shade tree selection by farmers impacts agroforests inter‐ and intraspecific trait diversity, and (iii) which shade tree traits should be considered for achieving farmer priorities. We consolidate literature on Coffea arabica (coffee) and Theobroma cacao (cocoa) agroforestry systems to summarize the role of shade tree functional traits in three key ecosystem functions: soil fertility, microclimate modification and crop productivity. We compile global and regional datasets on tree functional trait diversity to show the functional space of agroforestry tree species compared with the overall functional space observed in plants. Despite, or maybe because of, high shade tree diversity, shade tree trait characterization remains coarse and commonly measured at the community scale in the literature. Based on published trait data, we show that farmers adjust the functional composition of shade trees to increase the recycling of soil nutrients (high leaf nitrogen), the production of wood (skewing towards lower wood densities) and the production of fruits (tendency towards high seed size). Common shade trees in coffee and cocoa systems fall in the mid‐range of leaf acquisitive to conservative strategies, providing evidence that expanding shade tree portfolios can improve, or even accelerate, functions. Synthesis and applications: Based on the agroforestry literature and on trait‐environment relationships, we propose a matrix of shade tree traits that influence desirable agroecosystem outcomes for farmers, which can guide fine‐scale coordination of trait expression and agroforestry functions.