feat: add TriageServiceRequest tool and related graph for analyzing service requests and drafting tickets

internal/ai/runtime/graphs/triage_service_request_graph_test.go

@@ -0,0 +1,47 @@
+package graphs
+
+import (
+	"testing"
+
+	"cs-agent/internal/models"
+	"cs-agent/internal/pkg/enums"
+)
+
+func TestTriageServiceRequestResult_PrepareTicket(t *testing.T) {
+	conversation := &models.Conversation{
+		Subject:            "支付失败需要登记工单",
+		LastMessageSummary: "用户要求建单跟进支付失败问题",
+	}
+	messages := []models.Message{
+		{SenderType: enums.IMSenderTypeCustomer, Content: "帮我建个工单，支付一直失败"},
+	}
+
+	analysis := buildAnalyzeConversationResult(conversation, messages, AnalyzeConversationInput{
+		NeedTicket: true,
+	})
+	if analysis.RecommendedNextAction != "prepare_ticket" {
+		t.Fatalf("expected prepare_ticket, got %q", analysis.RecommendedNextAction)
+	}
+
+	draft := buildPrepareTicketDraftResult(conversation, messages, PrepareTicketDraftInput{})
+	if draft.Title == "" || draft.Description == "" {
+		t.Fatalf("expected draft to be populated, got %#v", draft)
+	}
+}
+
+func TestTriageServiceRequestResult_Handoff(t *testing.T) {
+	conversation := &models.Conversation{
+		Subject:            "投诉转人工",
+		LastMessageSummary: "用户要求人工处理扣费投诉",
+	}
+	messages := []models.Message{
+		{SenderType: enums.IMSenderTypeCustomer, Content: "我要投诉并转人工，你们重复扣费了"},
+	}
+
+	analysis := buildAnalyzeConversationResult(conversation, messages, AnalyzeConversationInput{
+		NeedHumanHandoff: true,
+	})
+	if analysis.RecommendedNextAction != "handoff_to_human" {
+		t.Fatalf("expected handoff_to_human, got %q", analysis.RecommendedNextAction)
+	}
+}