Direct delivery of therapeutic genes is a promising approach for treating cancers and other diseases. The current human viral vectors, however, suffer from several drawbacks, including poor cell-type specificity and difficult large-scale production. The M13 phage provides an alternative vehicle for gene therapy with engineerable specificity, but the low transduction efficiency seriously limits its translational application. In this work, we discovered important factors of cells and phages that greatly influence the phage transduction. The up-regulation of PrimPol or the down-regulation of DMBT1 in cells significantly enhanced the phage transduction efficiency. Furthermore, we found that the phage transduction efficiency was inversely correlated with the phage size. By carefully reconstructing the phage origin with the gene of interest, we designed "TransPhage" with a minimal length and maximal transduction efficiency. We showed that TransPhage successfully transduced the human cells with an excellent efficiency (up to 95%) comparable to or superior to that of the adeno-associated virus vectors. Moreover, we showed that TransPhage's tropism was specific to the cells that overexpress the target antigen, whereas adeno-associated viruses (AAVs) promiscuously infected many cell types. Using TransPhage as a gene therapy vehicle, we invented an NK-cell-mediated immunotherapy in which a membrane-bound fragment crystallizable region was introduced to cancer cells. We showed in vitro that the cancer cells expressing the membrane-bound fragment crystallizable (Fc) were effectively killed by CD16+ NK cells through an antibody-dependent cell-mediated cytotoxicity (ADCC)-like mechanism. In the xenograft mouse model, the administration of TransPhage carrying the membrane-bound Fc gene greatly suppressed tumor growth.