diff --git a/.github/FUNDING.yml b/.github/FUNDING.yml
new file mode 100644
index 0000000..3040112
--- /dev/null
+++ b/.github/FUNDING.yml
@@ -0,0 +1,2 @@
+github: hayabusa-cloud
+
diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000..df45b80
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,17 @@
+.PHONY: test bench cover vet clean
+
+test:
+	go test -race ./...
+
+bench:
+	go test -bench=. -benchmem ./...
+
+cover:
+	go test -coverprofile=coverage.out ./...
+	go tool cover -func=coverage.out | tail -1
+
+vet:
+	go vet ./...
+
+clean:
+	rm -f coverage.out
diff --git a/benchmark_test.go b/benchmark_test.go
new file mode 100644
index 0000000..379f0c3
--- /dev/null
+++ b/benchmark_test.go
@@ -0,0 +1,242 @@
+// ©Hayabusa Cloud Co., Ltd. 2026. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package spin_test
+
+import (
+	"runtime"
+	"sync"
+	"testing"
+
+	"code.hybscloud.com/spin"
+)
+
+// ----------------------------------------------------------------------------
+// Lock Benchmarks
+// ----------------------------------------------------------------------------
+
+// BenchmarkLock measures lock performance across contention levels.
+func BenchmarkLock(b *testing.B) {
+	b.Run("Uncontended", func(b *testing.B) {
+		var lk spin.Lock
+		for range b.N {
+			lk.Lock()
+			lk.Unlock()
+		}
+	})
+
+	b.Run("Try/Uncontended", func(b *testing.B) {
+		var lk spin.Lock
+		for range b.N {
+			if lk.Try() {
+				lk.Unlock()
+			}
+		}
+	})
+
+	// Contention scaling benchmarks
+	for _, procs := range []int{2, 4, 8, 16} {
+		b.Run("Contended/GOMAXPROCS="+itoa(procs), func(b *testing.B) {
+			oldProcs := runtime.GOMAXPROCS(procs)
+			defer runtime.GOMAXPROCS(oldProcs)
+
+			var lk spin.Lock
+			b.RunParallel(func(pb *testing.PB) {
+				for pb.Next() {
+					lk.Lock()
+					lk.Unlock()
+				}
+			})
+		})
+	}
+}
+
+// BenchmarkLockVsMutex compares spin.Lock against sync.Mutex.
+func BenchmarkLockVsMutex(b *testing.B) {
+	b.Run("spin.Lock/Uncontended", func(b *testing.B) {
+		var lk spin.Lock
+		for range b.N {
+			lk.Lock()
+			lk.Unlock()
+		}
+	})
+
+	b.Run("sync.Mutex/Uncontended", func(b *testing.B) {
+		var mu sync.Mutex
+		for range b.N {
+			mu.Lock()
+			mu.Unlock()
+		}
+	})
+
+	b.Run("spin.Lock/Contended", func(b *testing.B) {
+		var lk spin.Lock
+		b.RunParallel(func(pb *testing.PB) {
+			for pb.Next() {
+				lk.Lock()
+				lk.Unlock()
+			}
+		})
+	})
+
+	b.Run("sync.Mutex/Contended", func(b *testing.B) {
+		var mu sync.Mutex
+		b.RunParallel(func(pb *testing.PB) {
+			for pb.Next() {
+				mu.Lock()
+				mu.Unlock()
+			}
+		})
+	})
+}
+
+// BenchmarkLockWithWork measures lock performance with simulated work.
+func BenchmarkLockWithWork(b *testing.B) {
+	work := func() {
+		// Simulate ~10ns of work
+		x := 0
+		for i := range 10 {
+			x += i
+		}
+		_ = x
+	}
+
+	b.Run("spin.Lock", func(b *testing.B) {
+		var lk spin.Lock
+		b.RunParallel(func(pb *testing.PB) {
+			for pb.Next() {
+				lk.Lock()
+				work()
+				lk.Unlock()
+			}
+		})
+	})
+
+	b.Run("sync.Mutex", func(b *testing.B) {
+		var mu sync.Mutex
+		b.RunParallel(func(pb *testing.PB) {
+			for pb.Next() {
+				mu.Lock()
+				work()
+				mu.Unlock()
+			}
+		})
+	})
+}
+
+// ----------------------------------------------------------------------------
+// Wait Benchmarks
+// ----------------------------------------------------------------------------
+
+// BenchmarkWait measures Wait performance.
+func BenchmarkWait(b *testing.B) {
+	b.Run("Once/Fresh", func(b *testing.B) {
+		for range b.N {
+			var sw spin.Wait
+			sw.Once()
+		}
+	})
+
+	b.Run("Once/Repeated", func(b *testing.B) {
+		var sw spin.Wait
+		for range b.N {
+			sw.Once()
+		}
+	})
+
+	b.Run("WillYield", func(b *testing.B) {
+		var sw spin.Wait
+		for range b.N {
+			_ = sw.WillYield()
+		}
+	})
+
+	b.Run("Reset", func(b *testing.B) {
+		var sw spin.Wait
+		for i := range b.N {
+			if i%16 == 0 {
+				sw.Reset()
+			}
+			sw.Once()
+		}
+	})
+}
+
+// ----------------------------------------------------------------------------
+// Pause Benchmarks
+// ----------------------------------------------------------------------------
+
+// BenchmarkPauseCycles measures Pause across different cycle counts.
+func BenchmarkPauseCycles(b *testing.B) {
+	for _, cycles := range []int{1, 10, 30, 50, 100} {
+		b.Run("cycles="+itoa(cycles), func(b *testing.B) {
+			for range b.N {
+				spin.Pause(cycles)
+			}
+		})
+	}
+}
+
+// BenchmarkPauseLoop measures pause in a tight loop (common pattern).
+func BenchmarkPauseLoop(b *testing.B) {
+	b.Run("10_iterations", func(b *testing.B) {
+		for range b.N {
+			for range 10 {
+				spin.Pause()
+			}
+		}
+	})
+
+	b.Run("100_iterations", func(b *testing.B) {
+		for range b.N {
+			for range 100 {
+				spin.Pause()
+			}
+		}
+	})
+}
+
+// ----------------------------------------------------------------------------
+// Spin-Wait Pattern Benchmarks
+// ----------------------------------------------------------------------------
+
+// BenchmarkSpinWaitPattern measures realistic spin-wait scenarios.
+func BenchmarkSpinWaitPattern(b *testing.B) {
+	// Pattern 1: Spin until condition (Wait helper)
+	b.Run("Wait/UntilCondition", func(b *testing.B) {
+		for range b.N {
+			var sw spin.Wait
+			for range 16 {
+				sw.Once()
+			}
+		}
+	})
+
+	// Pattern 2: Raw pause loop
+	b.Run("Pause/RawLoop", func(b *testing.B) {
+		for range b.N {
+			for range 16 {
+				spin.Pause()
+			}
+		}
+	})
+}
+
+// ----------------------------------------------------------------------------
+// Helpers
+// ----------------------------------------------------------------------------
+
+func itoa(n int) string {
+	if n == 0 {
+		return "0"
+	}
+	var buf [20]byte
+	i := len(buf)
+	for n > 0 {
+		i--
+		buf[i] = byte('0' + n%10)
+		n /= 10
+	}
+	return string(buf[i:])
+}