diff --git a/Makefile b/Makefile
index f678886..045fd89 100644
--- a/Makefile
+++ b/Makefile
@@ -51,19 +51,11 @@ all: kern/kernel.elf
 quickstart:
 	make get_toolchain && bear -- make -j$(nproc) && make qemu
 
-KERNEL_OBJ := \
-	kern/entry.o \
-	kern/start.o \
-	kern/kalloc.o \
-	kern/libkern/string.o \
-	kern/libkern/proc.o \
-	kern/libkern/uart.o \
-	kern/libkern/panic.o \
-	kern/libkern/memory.o \
-	kern/libkern/spinlock.o \
-	kern/libkern/mini-printf.o \
-	kern/libkern/stdio.o \
-	kern/libkern/badrand.o
+SUBDIRS := kern
+
+include $(patsubst %, %/include.mk, $(SUBDIRS))
+
+KERNEL_OBJ := $(KERN_OBJ)
 
 kern/kernel.elf: $(KERNEL_OBJ)
 	@echo LD $@
diff --git a/README.md b/README.md
index 8d95e18..c7afa7d 100644
--- a/README.md
+++ b/README.md
@@ -26,6 +26,7 @@ On a linux machine, make sure that you have `make` and `curl` in path. The rest
 of the tooling (gcc, qemu) will be handled by the build system. For extra convenience,
 you could also use `bear` for generating "compile_commands.json". This is useful for helping
 clangd pick up the correct paths, and will result in better in-editor diagnostics and warnings.
+clang-format and find might also be useful for development.
 
 ```sh
 make get_toolchain # Dont worry, is will land locally inside project directory
diff --git a/kern/include.mk b/kern/include.mk
new file mode 100644
index 0000000..8496555
--- /dev/null
+++ b/kern/include.mk
@@ -0,0 +1,9 @@
+KERN_SRC  := $(wildcard kern/*.c)
+KERN_SRC += $(wildcard kern/*.S)
+
+KERN_OBJ  := $(KERN_SRC:.c=.o)
+KERN_OBJ  := $(KERN_OBJ:.S=.o)
+
+include kern/libkern/include.mk
+
+KERN_OBJ += $(LIBKERN_OBJ)
diff --git a/kern/libkern/include.mk b/kern/libkern/include.mk
new file mode 100644
index 0000000..491b247
--- /dev/null
+++ b/kern/libkern/include.mk
@@ -0,0 +1,5 @@
+LIBKERN_SRC := $(wildcard kern/libkern/*.c)
+LIBKERN_SRC += $(wildcard kern/libkern/*.S)
+
+LIBKERN_OBJ  := $(LIBKERN_SRC:.c=.o)
+LIBKERN_OBJ  := $(LIBKERN_OBJ:.S=.o)