fix: health check includes response body in errors, adds 503 test

Co-Authored-By: Virgil <virgil@lethean.io>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

internal/llamacpp/health.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
+	"io"
 	"net/http"
 	"strings"
 )
@@ -39,7 +40,8 @@ func (c *Client) Health(ctx context.Context) error {
 	defer resp.Body.Close()
 
 	if resp.StatusCode != http.StatusOK {
-		return fmt.Errorf("llamacpp: health returned %d", resp.StatusCode)
+		body, _ := io.ReadAll(io.LimitReader(resp.Body, 256))
+		return fmt.Errorf("llamacpp: health returned %d: %s", resp.StatusCode, string(body))
 	}
 	var h healthResponse
 	if err := json.NewDecoder(resp.Body).Decode(&h); err != nil {

internal/llamacpp/health_test.go

@@ -35,6 +35,19 @@ func TestHealth_NotReady(t *testing.T) {
 	assert.ErrorContains(t, err, "not ready")
 }
 
+func TestHealth_Loading(t *testing.T) {
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.Header().Set("Content-Type", "application/json")
+		w.WriteHeader(http.StatusServiceUnavailable)
+		w.Write([]byte(`{"error":{"code":503,"message":"Loading model","type":"unavailable_error"}}`))
+	}))
+	defer ts.Close()
+
+	c := NewClient(ts.URL)
+	err := c.Health(context.Background())
+	assert.ErrorContains(t, err, "503")
+}
+
 func TestHealth_ServerDown(t *testing.T) {
 	c := NewClient("http://127.0.0.1:1") // nothing listening
 	err := c.Health(context.Background())