Despite the tremendous progress made toward establishing PLS as a new paradigm to guarantee security of communication systems at the physical layerthere is a common belief among researchers and industrials that there are many practical challenges that prevent PLS from flourishing at the industrial scale. Most secure message transmission constructions available to date are tied to strong assumptions on CSI, consider simple channel models and undermine eavesdropping capabilities; thus compromising their practical interest to a big extent. Perhaps arguably, the most likely reasonable way to leverage PLS potential in securing modern wireless communication systems is via secret-key agreement. In the latter setting, the legitimate parties try to agree on a key exploiting availability of a public channel with high capacity which is also accessible to the eavesdropper. Once a key is shared by the legitimate parties, they may use it in a one-time pad encryption, for instance. In this article, we investigate two performance limits of secret-key agreement communications; namely, the secret-key diversity-multiplexing trade-off and the effect of transmit correlation on the secretkey capacity. We show via examples how secretkey agreement offers more flexibility than secure message transmissions. Finally, we explore a few challenges of secret-key agreement concept and propose a few guidelines to overturn them.