Smart　home　is　an　important　application　area　of　the　Internet　of　things　(IoT).　However,　the　diversification　of　smart　home　application　scenarios　increases　the　difficulty　of　understanding　the　scenarios　for　developers.　And　the　heterogeneity　of　the　programming　interfaces　of　smart　devices　as　well　as　the　close　coupling　of　the　code　to　the　underlying　systems　is　still　an　important　work　for　the　developers.　Furthermore,　the　complexity　and　variability　of　business　requirements　poses　a　great　challenge　to　the　development　of　applications　logic.　In　this　paper,　we　present　a　runtime　knowledge　graph　based　approach　to　smart　home　application　development.　First,　a　conceptual　model　describing　the　smart　home　scenarios　is　defined.　Second,　the　manageability　of　smart　devices　is　abstracted　as　runtime　knowledge　graphs　that　are　automatically　connected　with　the　corresponding　systems.　Last,　a　method　of　automatically　generating　smart　home　applications　is　proposed.　Our　approach　can　reduce　code　by　about　85　percent　at　least,　and　an　experiment　on　a　real-world　application　scenario　demonstrates　the　feasibility,　effectiveness,　and　benefits　of　the　new　approach　to　smart　home　application　development.