diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 2a94727..8718f8e 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -2,8 +2,8 @@ name: build
 
 on:
   push:
-    branches:
-    tags:
+    branches: ["**"]
+    tags: ["**"]
   pull_request:
 
 jobs:
@@ -12,10 +12,10 @@ jobs:
 
     steps:
       - name: checkout
-        uses: actions/checkout@v4
+        uses: actions/checkout@v7
 
       - name: set up go
-        uses: actions/setup-go@v5
+        uses: actions/setup-go@v7
         with:
           go-version: "1.24"
         id: go
@@ -29,9 +29,9 @@ jobs:
           go build -race
 
       - name: golangci-lint
-        uses: golangci/golangci-lint-action@v7
+        uses: golangci/golangci-lint-action@v9
         with:
-          version: v2.7
+          version: v2.12.2
 
       - name: install goveralls
         run: |
diff --git a/bench_test.go b/bench_test.go
index 3100678..ee66f90 100644
--- a/bench_test.go
+++ b/bench_test.go
@@ -36,7 +36,7 @@ func TestPoolPerf(t *testing.T) {
 
 	var egDuration time.Duration
 	t.Run("errgroup", func(t *testing.T) {
-		var count2 int32
+		var count2 atomic.Int32
 		st := time.Now()
 		defer func() {
 			egDuration = time.Since(st)
@@ -47,20 +47,20 @@ func TestPoolPerf(t *testing.T) {
 		for range 1000000 {
 			g.Go(func() error {
 				benchTask(n)
-				atomic.AddInt32(&count2, 1)
+				count2.Add(1)
 				return nil
 			})
 		}
 		require.NoError(t, g.Wait())
-		assert.Equal(t, int32(1000000), atomic.LoadInt32(&count2))
+		assert.Equal(t, int32(1000000), count2.Load())
 	})
 
 	t.Run("pool default", func(t *testing.T) {
 		// pool with 8 workers
-		var count1 int32
+		var count1 atomic.Int32
 		worker := WorkerFunc[int](func(context.Context, int) error {
 			benchTask(n)
-			atomic.AddInt32(&count1, 1)
+			count1.Add(1)
 			return nil
 		})
 
@@ -74,17 +74,17 @@ func TestPoolPerf(t *testing.T) {
 			assert.NoError(t, p.Close(ctx))
 		}()
 		require.NoError(t, p.Wait(ctx))
-		assert.Equal(t, int32(1000000), atomic.LoadInt32(&count1))
+		assert.Equal(t, int32(1000000), count1.Load())
 		t.Logf("elapsed pool: %v", time.Since(st))
 		assert.Less(t, time.Since(st), egDuration)
 	})
 
 	t.Run("pool with 100 chan size", func(t *testing.T) {
 		// pool with 8 workers
-		var count1 int32
+		var count1 atomic.Int32
 		worker := WorkerFunc[int](func(context.Context, int) error {
 			benchTask(n)
-			atomic.AddInt32(&count1, 1)
+			count1.Add(1)
 			return nil
 		})
 
@@ -98,17 +98,17 @@ func TestPoolPerf(t *testing.T) {
 			assert.NoError(t, p.Close(ctx))
 		}()
 		require.NoError(t, p.Wait(ctx))
-		assert.Equal(t, int32(1000000), atomic.LoadInt32(&count1))
+		assert.Equal(t, int32(1000000), count1.Load())
 		t.Logf("elapsed pool: %v", time.Since(st))
 		assert.Less(t, time.Since(st), egDuration)
 	})
 
 	t.Run("pool with 100 chan size and 100 batch size", func(t *testing.T) {
 		// pool with 8 workers
-		var count1 int32
+		var count1 atomic.Int32
 		worker := WorkerFunc[int](func(context.Context, int) error {
 			benchTask(n)
-			atomic.AddInt32(&count1, 1)
+			count1.Add(1)
 			return nil
 		})
 
@@ -122,17 +122,17 @@ func TestPoolPerf(t *testing.T) {
 			assert.NoError(t, p.Close(ctx))
 		}()
 		require.NoError(t, p.Wait(ctx))
-		assert.Equal(t, int32(1000000), atomic.LoadInt32(&count1))
+		assert.Equal(t, int32(1000000), count1.Load())
 		t.Logf("elapsed pool: %v", time.Since(st))
 		assert.Less(t, time.Since(st), egDuration)
 	})
 
 	t.Run("pool with 100 chan size and 100 batch size and chunking", func(t *testing.T) {
 		// pool with 8 workers
-		var count1 int32
+		var count1 atomic.Int32
 		worker := WorkerFunc[int](func(context.Context, int) error {
 			benchTask(n)
-			atomic.AddInt32(&count1, 1)
+			count1.Add(1)
 			return nil
 		})
 
@@ -148,7 +148,7 @@ func TestPoolPerf(t *testing.T) {
 			assert.NoError(t, p.Close(ctx))
 		}()
 		require.NoError(t, p.Wait(ctx))
-		assert.Equal(t, int32(1000000), atomic.LoadInt32(&count1))
+		assert.Equal(t, int32(1000000), count1.Load())
 		t.Logf("elapsed pool: %v", time.Since(st))
 		assert.Less(t, time.Since(st), egDuration)
 	})
@@ -164,29 +164,29 @@ func BenchmarkPoolCompare(b *testing.B) {
 	b.Run("errgroup", func(b *testing.B) {
 		b.ResetTimer()
 		for range b.N {
-			var count int32
+			var count atomic.Int32
 			g, _ := errgroup.WithContext(ctx)
 			g.SetLimit(workers)
 
 			for range iterations {
 				g.Go(func() error {
 					benchTask(n)
-					atomic.AddInt32(&count, 1)
+					count.Add(1)
 					return nil
 				})
 			}
 			require.NoError(b, g.Wait())
-			require.Equal(b, int32(iterations), atomic.LoadInt32(&count))
+			require.Equal(b, int32(iterations), count.Load())
 		}
 	})
 
 	b.Run("pool default", func(b *testing.B) {
 		b.ResetTimer()
 		for range b.N {
-			var count int32
+			var count atomic.Int32
 			p := New[int](workers, WorkerFunc[int](func(context.Context, int) error {
 				benchTask(n)
-				atomic.AddInt32(&count, 1)
+				count.Add(1)
 				return nil
 			}))
 
@@ -198,17 +198,17 @@ func BenchmarkPoolCompare(b *testing.B) {
 				p.Close(ctx)
 			}()
 			require.NoError(b, p.Wait(ctx))
-			require.Equal(b, int32(iterations), atomic.LoadInt32(&count))
+			require.Equal(b, int32(iterations), count.Load())
 		}
 	})
 
 	b.Run("pool with chan=100", func(b *testing.B) {
 		b.ResetTimer()
 		for range b.N {
-			var count int32
+			var count atomic.Int32
 			p := New[int](workers, WorkerFunc[int](func(context.Context, int) error {
 				benchTask(n)
-				atomic.AddInt32(&count, 1)
+				count.Add(1)
 				return nil
 			})).WithWorkerChanSize(100)
 
@@ -220,17 +220,17 @@ func BenchmarkPoolCompare(b *testing.B) {
 				p.Close(ctx)
 			}()
 			require.NoError(b, p.Wait(ctx))
-			require.Equal(b, int32(iterations), atomic.LoadInt32(&count))
+			require.Equal(b, int32(iterations), count.Load())
 		}
 	})
 
 	b.Run("pool with batching", func(b *testing.B) {
 		b.ResetTimer()
 		for range b.N {
-			var count int32
+			var count atomic.Int32
 			p := New[int](workers, WorkerFunc[int](func(context.Context, int) error {
 				benchTask(n)
-				atomic.AddInt32(&count, 1)
+				count.Add(1)
 				return nil
 			})).WithWorkerChanSize(100).WithBatchSize(100)
 
@@ -242,17 +242,17 @@ func BenchmarkPoolCompare(b *testing.B) {
 				p.Close(ctx)
 			}()
 			require.NoError(b, p.Wait(ctx))
-			require.Equal(b, int32(iterations), atomic.LoadInt32(&count))
+			require.Equal(b, int32(iterations), count.Load())
 		}
 	})
 
 	b.Run("pool with batching and chunking", func(b *testing.B) {
 		b.ResetTimer()
 		for range b.N {
-			var count int32
+			var count atomic.Int32
 			p := New[int](workers, WorkerFunc[int](func(context.Context, int) error {
 				benchTask(n)
-				atomic.AddInt32(&count, 1)
+				count.Add(1)
 				return nil
 			})).WithWorkerChanSize(100).WithBatchSize(100).WithChunkFn(func(v int) string {
 				return strconv.Itoa(v % workers)
@@ -266,7 +266,7 @@ func BenchmarkPoolCompare(b *testing.B) {
 				p.Close(ctx)
 			}()
 			require.NoError(b, p.Wait(ctx))
-			require.Equal(b, int32(iterations), atomic.LoadInt32(&count))
+			require.Equal(b, int32(iterations), count.Load())
 		}
 	})
 }
diff --git a/pool.go b/pool.go
index 0e9ded7..8aea9c1 100644
--- a/pool.go
+++ b/pool.go
@@ -6,6 +6,7 @@ import (
 	"fmt"
 	"hash/fnv"
 	"math/rand"
+	"slices"
 	"sync"
 	"sync/atomic"
 	"time"
@@ -566,9 +567,9 @@ func (p *WorkerGroup[T]) Use(middlewares ...Middleware[T]) *WorkerGroup[T] {
 			// apply middlewares in order from last to first
 			// this makes first middleware outermost
 			wrapped := worker
-			for i := len(middlewares) - 1; i >= 0; i-- {
+			for _, v := range slices.Backward(middlewares) {
 				prev := wrapped
-				wrapped = middlewares[i](prev)
+				wrapped = v(prev)
 			}
 			return wrapped
 		}
@@ -577,9 +578,9 @@ func (p *WorkerGroup[T]) Use(middlewares ...Middleware[T]) *WorkerGroup[T] {
 
 	// for stateless worker, just wrap it directly
 	wrapped := p.worker
-	for i := len(middlewares) - 1; i >= 0; i-- {
+	for _, v := range slices.Backward(middlewares) {
 		prev := wrapped
-		wrapped = middlewares[i](prev)
+		wrapped = v(prev)
 	}
 	p.worker = wrapped
 	return p
diff --git a/pool_test.go b/pool_test.go
index 1d8ca62..994b152 100644
--- a/pool_test.go
+++ b/pool_test.go
@@ -333,10 +333,10 @@ func TestPool_Distribution(t *testing.T) {
 
 func TestPool_Metrics(t *testing.T) {
 	t.Run("basic metrics", func(t *testing.T) {
-		var processed int32
+		var processed atomic.Int32
 		worker := WorkerFunc[int](func(ctx context.Context, _ int) error {
 			time.Sleep(time.Millisecond) // simulate work
-			atomic.AddInt32(&processed, 1)
+			processed.Add(1)
 			return nil
 		})
 
@@ -349,7 +349,7 @@ func TestPool_Metrics(t *testing.T) {
 		require.NoError(t, p.Close(context.Background()))
 
 		stats := p.Metrics().GetStats()
-		assert.Equal(t, int(atomic.LoadInt32(&processed)), stats.Processed)
+		assert.Equal(t, int(processed.Load()), stats.Processed)
 		assert.Equal(t, 0, stats.Errors)
 		assert.Equal(t, 0, stats.Dropped)
 		assert.Greater(t, stats.ProcessingTime, time.Duration(0))
@@ -948,12 +948,12 @@ func TestMiddleware_Practical(t *testing.T) {
 			return nil
 		})
 
-		var totalTime int64
+		var totalTime atomic.Int64
 		timingMiddleware := func(next Worker[string]) Worker[string] {
 			return WorkerFunc[string](func(ctx context.Context, v string) error {
 				start := time.Now()
 				err := next.Do(ctx, v)
-				atomic.AddInt64(&totalTime, time.Since(start).Microseconds())
+				totalTime.Add(time.Since(start).Microseconds())
 				return err
 			})
 		}
@@ -964,7 +964,7 @@ func TestMiddleware_Practical(t *testing.T) {
 		p.Submit("test")
 		require.NoError(t, p.Close(context.Background()))
 
-		assert.Greater(t, atomic.LoadInt64(&totalTime), int64(1000),
+		assert.Greater(t, totalTime.Load(), int64(1000),
 			"should measure time greater than 1ms")
 	})
 }