Multiple agents in the family Filoviridae (filoviruses) are associated with sporadic human outbreaks of highly lethal disease, while others, including several recently identified agents, possess strong zoonotic potential. Although viral glycoprotein (GP)-specific monoclonal antibodies have demonstrated therapeutic utility against filovirus disease, currently FDA-approved molecules lack antiviral breadth. The development of broadly neutralizing antibodies has been challenged by the high sequence divergence among filovirus GPs and the complex GP proteolytic cleavage cascade that accompanies filovirus entry. Despite this variability in the antigenic surface of GP, all filoviruses share a site of vulnerability—the binding site for the universal filovirus entry receptor, Niemann-Pick C1 (NPC1). Unfortunately, this site is shielded in extracellular GP and only uncovered by proteolytic cleavage by host proteases in late endosomes and lysosomes, which are generally inaccessible to antibodies. To overcome this obstacle, we previously developed a ‘Trojan horse’ therapeutic approach in which engineered bispecific antibodies (bsAbs) coopt viral particles to deliver GP:NPC1 interaction-blocking antibodies to their endo/lysosomal sites of action. This approach afforded broad protection against members of the genus Ebolavirus but could not neutralize more divergent filoviruses. Here, we describe next-generation Trojan horse bsAbs that target the endo/lysosomal GP:NPC1 interface with pan-filovirus breadth by exploiting the conserved and widely expressed host cation-independent mannose-6-phosphate receptor for intracellular delivery. Our work highlights a new avenue for the development of single therapeutics protecting against all known and newly emerging filoviruses.