Experimental investigations have been increasingly conducted into intervention techniques for strengthening traditional masonry elements by a combination of corrosion-inert FRP rebars and a ductile (cement-based or adhesive) embedding matrix. The design of such interventions depends critically on the bond properties between rebar and matrix. In this work, an alternative setup for the direct-tension test is introduced, which offers increased reliability and repeatability. In this setup, a double-T specimen is incorporated in a specially-designed rombus-shaped metal apparatus as the diagonal tie in tension, and the assemblage is subjected to compression. The proposed setup may be employed to characterize the behaviour of a strain resilient cementitious composite (SRCC) in direct tension as well as the local bond between itself and embedded FRP rebars. In the latter case the specimen features a pre-formed gap in the matrix (bridged by the rebar), dividing said specimen into a short “study” length and a longer “support” length. Under direct tension, the rebar slips relative to the matrix, the gap widens, and local bond property values are computed from load and displacement measurements. These values are compared to those obtained from bending of beam specimens (an implicit tension test known to yield results sensitive to curvature).