feat(multimodal): 视觉分层路由（多模态检测→主/视觉/OCR 工具）+ 语音 STT→TTS 降级链（异步）

internal/runtime/multimodal/vision/adapter_test.go

@@ -0,0 +1,53 @@
+package vision
+
+import (
+	"context"
+	"encoding/json"
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"git.yeij.top/AskaEth/Cyrene_v2/internal/config"
+)
+
+func TestOpenAICompatibleAdapter_Describe(t *testing.T) {
+	t.Parallel()
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.URL.Path != "/chat/completions" {
+			t.Errorf("path = %s, want /chat/completions", r.URL.Path)
+		}
+		var body map[string]any
+		if err := json.NewDecoder(r.Body).Decode(&body); err != nil {
+			t.Fatalf("decode body: %v", err)
+		}
+		msgs, _ := body["messages"].([]any)
+		if len(msgs) != 1 {
+			t.Fatalf("messages = %d, want 1", len(msgs))
+		}
+		content, _ := msgs[0].(map[string]any)["content"].([]any)
+		if len(content) != 2 {
+			t.Fatalf("content parts = %d, want 2", len(content))
+		}
+		w.Header().Set("Content-Type", "application/json")
+		_, _ = io.WriteString(w, `{"choices":[{"message":{"content":"a red car"}}]}`)
+	}))
+	defer srv.Close()
+
+	ad := NewOpenAICompatible(config.LLMConfig{BaseURL: srv.URL, Model: "vision-"})
+	text, err := ad.Describe(context.Background(), ImageRef{Source: SourceBase64, Data: "aGVsbG8="}, "")
+	if err != nil {
+		t.Fatalf("Describe error: %v", err)
+	}
+	if text != "a red car" {
+		t.Errorf("text = %q, want a red car", text)
+	}
+}
+
+func TestOpenAICompatibleAdapter_NotConfigured(t *testing.T) {
+	t.Parallel()
+	ad := NewOpenAICompatible(config.LLMConfig{})
+	if _, err := ad.Describe(context.Background(), ImageRef{Source: SourceBase64, Data: "aGVsbG8="}, ""); err == nil {
+		t.Fatal("want error for empty base url, got nil")
+	}
+}