diff --git a/ingest_test.go b/ingest_test.go
new file mode 100644
index 0000000..81b2369
--- /dev/null
+++ b/ingest_test.go
@@ -0,0 +1,391 @@
+package quickwit_test
+
+import (
+	"encoding/json"
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/moonrhythm/quickwit"
+)
+
+// parseIndices extracts the "index" field of every NDJSON line in an ingest body.
+func parseIndices(body []byte) []int {
+	var out []int
+	for _, line := range strings.Split(strings.TrimSpace(string(body)), "\n") {
+		if line == "" {
+			continue
+		}
+		var m map[string]any
+		if json.Unmarshal([]byte(line), &m) != nil {
+			continue
+		}
+		if v, ok := m["index"].(float64); ok {
+			out = append(out, int(v))
+		}
+	}
+	return out
+}
+
+// Core: every ingested record reaches the server exactly once, in order
+// (single worker), across multiple batches.
+func TestIngest_DeliversAllItemsInOrder(t *testing.T) {
+	const numItems = 250
+
+	var mu sync.Mutex
+	var received []int
+
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		body, _ := io.ReadAll(r.Body)
+		mu.Lock()
+		received = append(received, parseIndices(body)...)
+		mu.Unlock()
+		w.WriteHeader(http.StatusOK)
+	}))
+	defer server.Close()
+
+	c := quickwit.NewClient(server.URL + "/api/v1/test")
+	c.SetConcurrent(1)
+	c.SetBatchSize(50)
+
+	for i := 0; i < numItems; i++ {
+		c.Ingest(map[string]any{"index": i})
+	}
+	c.Close()
+
+	mu.Lock()
+	defer mu.Unlock()
+
+	if len(received) != numItems {
+		t.Fatalf("received %d items, want %d", len(received), numItems)
+	}
+	for i, idx := range received {
+		if idx != i {
+			t.Errorf("position %d: got index %d, want %d", i, idx, i)
+		}
+	}
+}
+
+// Core: reaching the configured batch size flushes immediately, without waiting
+// for the timer. maxDelay is set far in the future so only the size trigger can fire.
+func TestIngest_FlushTriggeredByBatchSize(t *testing.T) {
+	const batchSize = 5
+
+	gotLines := make(chan int, 4)
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		body, _ := io.ReadAll(r.Body)
+		gotLines <- len(parseIndices(body))
+		w.WriteHeader(http.StatusOK)
+	}))
+	defer server.Close()
+
+	c := quickwit.NewClient(server.URL + "/api/v1/test")
+	c.SetConcurrent(1)
+	c.SetBatchSize(batchSize)
+	c.SetMaxDelay(10 * time.Second) // timer must not be the cause
+	defer c.Close()
+
+	for i := 0; i < batchSize; i++ {
+		c.Ingest(map[string]any{"index": i})
+	}
+
+	select {
+	case n := <-gotLines:
+		if n != batchSize {
+			t.Errorf("flushed %d records, want %d", n, batchSize)
+		}
+	case <-time.After(2 * time.Second):
+		t.Fatal("batch-size flush did not occur before timeout")
+	}
+}
+
+// Core: a partial batch flushes once the timer fires, even though the batch size
+// is never reached.
+func TestIngest_FlushTriggeredByTimer(t *testing.T) {
+	gotReq := make(chan struct{}, 4)
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		io.Copy(io.Discard, r.Body)
+		gotReq <- struct{}{}
+		w.WriteHeader(http.StatusOK)
+	}))
+	defer server.Close()
+
+	c := quickwit.NewClient(server.URL + "/api/v1/test")
+	c.SetConcurrent(1)
+	c.SetBatchSize(1000) // size trigger must not fire
+	c.SetMaxDelay(50 * time.Millisecond)
+	defer c.Close()
+
+	c.Ingest(
+		map[string]any{"index": 0},
+		map[string]any{"index": 1},
+		map[string]any{"index": 2},
+	)
+
+	select {
+	case <-gotReq:
+	case <-time.After(2 * time.Second):
+		t.Fatal("timer flush did not occur before timeout")
+	}
+}
+
+// Core: Close() flushes whatever remains buffered, even when neither the size nor
+// timer trigger has fired.
+func TestIngest_FlushRemainingOnClose(t *testing.T) {
+	var mu sync.Mutex
+	var received []int
+
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		body, _ := io.ReadAll(r.Body)
+		mu.Lock()
+		received = append(received, parseIndices(body)...)
+		mu.Unlock()
+		w.WriteHeader(http.StatusOK)
+	}))
+	defer server.Close()
+
+	c := quickwit.NewClient(server.URL + "/api/v1/test")
+	c.SetConcurrent(1)
+	c.SetBatchSize(1000)            // never reached
+	c.SetMaxDelay(10 * time.Second) // never fires before Close
+	c.Ingest(
+		map[string]any{"index": 0},
+		map[string]any{"index": 1},
+		map[string]any{"index": 2},
+	)
+	c.Close()
+
+	mu.Lock()
+	defer mu.Unlock()
+	if len(received) != 3 {
+		t.Fatalf("received %d items after Close, want 3", len(received))
+	}
+	for i, idx := range received {
+		if idx != i {
+			t.Errorf("position %d: got index %d, want %d", i, idx, i)
+		}
+	}
+}
+
+// Core: a failing flush retries (during normal operation, indefinitely) until the
+// server accepts it; the record must not be lost.
+func TestIngest_RetriesFailedFlushUntilSuccess(t *testing.T) {
+	var mu sync.Mutex
+	var attempts int
+	var delivered []int
+
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		body, _ := io.ReadAll(r.Body)
+		mu.Lock()
+		attempts++
+		a := attempts
+		if a >= 3 {
+			delivered = append(delivered, parseIndices(body)...)
+		}
+		mu.Unlock()
+		if a < 3 {
+			w.WriteHeader(http.StatusInternalServerError)
+			return
+		}
+		w.WriteHeader(http.StatusOK)
+	}))
+	defer server.Close()
+
+	c := quickwit.NewClient(server.URL + "/api/v1/test")
+	c.SetConcurrent(1)
+	c.SetBatchSize(1) // each item triggers a flush
+	c.Ingest(map[string]any{"index": 42})
+	c.Close()
+
+	mu.Lock()
+	defer mu.Unlock()
+	if attempts < 3 {
+		t.Fatalf("server saw %d attempts, want at least 3 (2 failures then success)", attempts)
+	}
+	if len(delivered) != 1 || delivered[0] != 42 {
+		t.Errorf("delivered = %v, want [42]", delivered)
+	}
+}
+
+// Core: the user-supplied auth hook is applied to every ingest request.
+func TestIngest_AppliesAuthHook(t *testing.T) {
+	var mu sync.Mutex
+	var gotAuth string
+
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		mu.Lock()
+		gotAuth = r.Header.Get("Authorization")
+		mu.Unlock()
+		io.Copy(io.Discard, r.Body)
+		w.WriteHeader(http.StatusOK)
+	}))
+	defer server.Close()
+
+	c := quickwit.NewClient(server.URL + "/api/v1/test")
+	c.SetConcurrent(1)
+	c.SetAuth(func(req *http.Request) {
+		req.Header.Set("Authorization", "Bearer test-token")
+	})
+	c.Ingest(map[string]any{"index": 0})
+	c.Close()
+
+	mu.Lock()
+	defer mu.Unlock()
+	if gotAuth != "Bearer test-token" {
+		t.Errorf("Authorization header = %q, want %q", gotAuth, "Bearer test-token")
+	}
+}
+
+// Core: with multiple concurrent workers, every record is delivered exactly once
+// (no loss, no duplication).
+func TestIngest_ConcurrentWorkersDeliverEachItemOnce(t *testing.T) {
+	const numItems = 1000
+
+	var mu sync.Mutex
+	counts := make(map[int]int)
+
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		body, _ := io.ReadAll(r.Body)
+		idxs := parseIndices(body)
+		mu.Lock()
+		for _, idx := range idxs {
+			counts[idx]++
+		}
+		mu.Unlock()
+		w.WriteHeader(http.StatusOK)
+	}))
+	defer server.Close()
+
+	c := quickwit.NewClient(server.URL + "/api/v1/test")
+	c.SetConcurrent(4)
+	c.SetBatchSize(10)
+	for i := 0; i < numItems; i++ {
+		c.Ingest(map[string]any{"index": i})
+	}
+	c.Close()
+
+	mu.Lock()
+	defer mu.Unlock()
+	if len(counts) != numItems {
+		t.Fatalf("delivered %d distinct items, want %d", len(counts), numItems)
+	}
+	for i := 0; i < numItems; i++ {
+		if counts[i] != 1 {
+			t.Errorf("index %d delivered %d times, want 1", i, counts[i])
+		}
+	}
+}
+
+// Core: in discard mode, items that arrive while the buffer is full are dropped
+// and surfaced through OnDiscard instead of blocking the caller. The server is held
+// so the single worker stays blocked mid-flush and cannot drain the buffer.
+func TestIngest_DiscardsWhenBufferFull(t *testing.T) {
+	const bufferSize = 2
+	const numItems = 200
+
+	release := make(chan struct{})
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		<-release // block the worker mid-flush
+		io.Copy(io.Discard, r.Body)
+		w.WriteHeader(http.StatusOK)
+	}))
+	defer server.Close()
+
+	var mu sync.Mutex
+	var discarded int
+
+	c := quickwit.NewClient(server.URL + "/api/v1/test")
+	c.SetConcurrent(1)
+	c.SetBatchSize(1) // first item triggers the (blocking) flush
+	c.SetIngestBufferSize(bufferSize)
+	c.SetDiscard(true)
+	c.OnDiscard(func(any) {
+		mu.Lock()
+		discarded++
+		mu.Unlock()
+	})
+
+	// Each Ingest does a non-blocking send; OnDiscard fires synchronously here.
+	for i := 0; i < numItems; i++ {
+		c.Ingest(map[string]any{"index": i})
+	}
+
+	mu.Lock()
+	n := discarded
+	mu.Unlock()
+
+	// At most bufferSize items sit in the channel plus the one the worker pulled
+	// before blocking, so the overwhelming majority must have been discarded.
+	if n < numItems-bufferSize-2 {
+		t.Errorf("discarded %d items, want at least %d", n, numItems-bufferSize-2)
+	}
+
+	close(release) // let the worker finish so Close can return
+	c.Close()
+}
+
+// Core: a 413 response with auto-reduce enabled re-sends the buffer in smaller
+// chunks while still delivering every record in order.
+func TestIngest_413ReducesBatchSizeAndChunks(t *testing.T) {
+	const batchSize = 10
+	const numItems = batchSize
+
+	var mu sync.Mutex
+	var requestCount int
+	var successLineCounts []int
+	var received []int
+
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		body, _ := io.ReadAll(r.Body)
+		mu.Lock()
+		requestCount++
+		first := requestCount == 1
+		if !first {
+			idxs := parseIndices(body)
+			successLineCounts = append(successLineCounts, len(idxs))
+			received = append(received, idxs...)
+		}
+		mu.Unlock()
+
+		if first {
+			w.WriteHeader(http.StatusRequestEntityTooLarge)
+			return
+		}
+		w.WriteHeader(http.StatusOK)
+	}))
+	defer server.Close()
+
+	c := quickwit.NewClient(server.URL + "/api/v1/test")
+	c.SetConcurrent(1)
+	c.SetAutoReduceBatchSize(true)
+	c.SetBatchSize(batchSize)
+	c.SetMaxDelay(10 * time.Second)
+	for i := 0; i < numItems; i++ {
+		c.Ingest(map[string]any{"index": i})
+	}
+	c.Close()
+
+	mu.Lock()
+	defer mu.Unlock()
+
+	if requestCount < 3 {
+		t.Fatalf("requestCount = %d, want at least 3 (one 413, then smaller chunks)", requestCount)
+	}
+	for _, n := range successLineCounts {
+		if n >= batchSize {
+			t.Errorf("a post-413 request carried %d records, want fewer than %d", n, batchSize)
+		}
+	}
+	if len(received) != numItems {
+		t.Fatalf("received %d items, want %d", len(received), numItems)
+	}
+	for i, idx := range received {
+		if idx != i {
+			t.Errorf("position %d: got index %d, want %d", i, idx, i)
+		}
+	}
+}