fix: replace workflow hardcoding chat model for knowledge recall (#233)

backend/api/handler/coze/workflow_service_test.go

@@ -239,7 +239,7 @@ func newWfTestRunner(t *testing.T) *wfTestRunner {
 
 	mockTos := storageMock.NewMockStorage(ctrl)
 	mockTos.EXPECT().GetObjectUrl(gomock.Any(), gomock.Any(), gomock.Any()).Return("", nil).AnyTimes()
-	workflowRepo := service.NewWorkflowRepository(mockIDGen, db, redisClient, mockTos, cpStore)
+	workflowRepo := service.NewWorkflowRepository(mockIDGen, db, redisClient, mockTos, cpStore, nil)
 	mockey.Mock(appworkflow.GetWorkflowDomainSVC).Return(service.NewWorkflowService(workflowRepo)).Build()
 	mockey.Mock(workflow2.GetRepository).Return(workflowRepo).Build()
 

backend/application/application.go

@@ -190,8 +190,11 @@ func initPrimaryServices(ctx context.Context, basicServices *basicServices) (*pr
 		return nil, err
 	}
 
-	workflowDomainSVC := workflow.InitService(
+	workflowDomainSVC, err := workflow.InitService(ctx,
 		basicServices.toWorkflowServiceComponents(pluginSVC, memorySVC, knowledgeSVC))
+	if err != nil {
+		return nil, err
+	}
 
 	shortcutSVC := shortcutcmd.InitService(basicServices.infra.DB, basicServices.infra.IDGenSVC)
 

backend/application/internal/env.go

@@ -0,0 +1,108 @@
+/*
+ * Copyright 2025 coze-dev Authors
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package internal
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"strconv"
+
+	ao "github.com/cloudwego/eino-ext/components/model/ark"
+	"github.com/cloudwego/eino-ext/components/model/deepseek"
+	"github.com/cloudwego/eino-ext/components/model/gemini"
+	"github.com/cloudwego/eino-ext/components/model/ollama"
+	mo "github.com/cloudwego/eino-ext/components/model/openai"
+	"github.com/cloudwego/eino-ext/components/model/qwen"
+	"github.com/coze-dev/coze-studio/backend/infra/contract/chatmodel"
+	"google.golang.org/genai"
+)
+
+func GetBuiltinChatModel(ctx context.Context, envPrefix string) (bcm chatmodel.BaseChatModel, configured bool, err error) {
+	getEnv := func(key string) string {
+		if val := os.Getenv(envPrefix + key); val != "" {
+			return val
+		}
+		return os.Getenv(key)
+	}
+
+	switch getEnv("BUILTIN_CM_TYPE") {
+	case "openai":
+		byAzure, _ := strconv.ParseBool(getEnv("BUILTIN_CM_OPENAI_BY_AZURE"))
+		bcm, err = mo.NewChatModel(ctx, &mo.ChatModelConfig{
+			APIKey:  getEnv("BUILTIN_CM_OPENAI_API_KEY"),
+			ByAzure: byAzure,
+			BaseURL: getEnv("BUILTIN_CM_OPENAI_BASE_URL"),
+			Model:   getEnv("BUILTIN_CM_OPENAI_MODEL"),
+		})
+	case "ark":
+		bcm, err = ao.NewChatModel(ctx, &ao.ChatModelConfig{
+			APIKey:  getEnv("BUILTIN_CM_ARK_API_KEY"),
+			Model:   getEnv("BUILTIN_CM_ARK_MODEL"),
+			BaseURL: getEnv("BUILTIN_CM_ARK_BASE_URL"),
+		})
+	case "deepseek":
+		bcm, err = deepseek.NewChatModel(ctx, &deepseek.ChatModelConfig{
+			APIKey:  getEnv("BUILTIN_CM_DEEPSEEK_API_KEY"),
+			BaseURL: getEnv("BUILTIN_CM_DEEPSEEK_BASE_URL"),
+			Model:   getEnv("BUILTIN_CM_DEEPSEEK_MODEL"),
+		})
+	case "ollama":
+		bcm, err = ollama.NewChatModel(ctx, &ollama.ChatModelConfig{
+			BaseURL: getEnv("BUILTIN_CM_OLLAMA_BASE_URL"),
+			Model:   getEnv("BUILTIN_CM_OLLAMA_MODEL"),
+		})
+	case "qwen":
+		bcm, err = qwen.NewChatModel(ctx, &qwen.ChatModelConfig{
+			APIKey:  getEnv("BUILTIN_CM_QWEN_API_KEY"),
+			BaseURL: getEnv("BUILTIN_CM_QWEN_BASE_URL"),
+			Model:   getEnv("BUILTIN_CM_QWEN_MODEL"),
+		})
+	case "gemini":
+		backend, convErr := strconv.ParseInt(getEnv("BUILTIN_CM_GEMINI_BACKEND"), 10, 64)
+		if convErr != nil {
+			return nil, false, convErr
+		}
+		c, clientErr := genai.NewClient(ctx, &genai.ClientConfig{
+			APIKey:   getEnv("BUILTIN_CM_GEMINI_API_KEY"),
+			Backend:  genai.Backend(backend),
+			Project:  getEnv("BUILTIN_CM_GEMINI_PROJECT"),
+			Location: getEnv("BUILTIN_CM_GEMINI_LOCATION"),
+			HTTPOptions: genai.HTTPOptions{
+				BaseURL: getEnv("BUILTIN_CM_GEMINI_BASE_URL"),
+			},
+		})
+		if clientErr != nil {
+			return nil, false, clientErr
+		}
+		bcm, err = gemini.NewChatModel(ctx, &gemini.Config{
+			Client: c,
+			Model:  getEnv("BUILTIN_CM_GEMINI_MODEL"),
+		})
+	default:
+		// accept builtin chat model not configured
+	}
+
+	if err != nil {
+		return nil, false, fmt.Errorf("knowledge init openai chat mode failed, %w", err)
+	}
+	if bcm != nil {
+		configured = true
+	}
+
+	return
+}

backend/application/knowledge/init.go

@@ -28,25 +28,18 @@ import (
 	"github.com/cloudwego/eino-ext/components/embedding/ark"
 	ollamaEmb "github.com/cloudwego/eino-ext/components/embedding/ollama"
 	"github.com/cloudwego/eino-ext/components/embedding/openai"
-	ao "github.com/cloudwego/eino-ext/components/model/ark"
-	"github.com/cloudwego/eino-ext/components/model/deepseek"
-	"github.com/cloudwego/eino-ext/components/model/gemini"
-	"github.com/cloudwego/eino-ext/components/model/ollama"
-	mo "github.com/cloudwego/eino-ext/components/model/openai"
-	"github.com/cloudwego/eino-ext/components/model/qwen"
 	"github.com/cloudwego/eino/components/prompt"
 	"github.com/cloudwego/eino/schema"
+	"github.com/coze-dev/coze-studio/backend/application/internal"
 	"github.com/coze-dev/coze-studio/backend/infra/impl/embedding/http"
 	"github.com/milvus-io/milvus/client/v2/milvusclient"
 	"github.com/volcengine/volc-sdk-golang/service/vikingdb"
 	"github.com/volcengine/volc-sdk-golang/service/visual"
-	"google.golang.org/genai"
 	"gorm.io/gorm"
 
 	"github.com/coze-dev/coze-studio/backend/application/search"
 	knowledgeImpl "github.com/coze-dev/coze-studio/backend/domain/knowledge/service"
 	"github.com/coze-dev/coze-studio/backend/infra/contract/cache"
-	"github.com/coze-dev/coze-studio/backend/infra/contract/chatmodel"
 	"github.com/coze-dev/coze-studio/backend/infra/contract/document/nl2sql"
 	"github.com/coze-dev/coze-studio/backend/infra/contract/document/ocr"
 	"github.com/coze-dev/coze-studio/backend/infra/contract/document/searchstore"
@@ -131,7 +124,7 @@ func InitService(c *ServiceComponents) (*KnowledgeApplicationService, error) {
 	}
 
 	var rewriter messages2query.MessagesToQuery
-	if rewriterChatModel, _, err := getBuiltinChatModel(ctx, "M2Q_"); err != nil {
+	if rewriterChatModel, _, err := internal.GetBuiltinChatModel(ctx, "M2Q_"); err != nil {
 		return nil, err
 	} else {
 		filePath := filepath.Join(root, "resources/conf/prompt/messages_to_query_template_jinja2.json")
@@ -146,7 +139,7 @@ func InitService(c *ServiceComponents) (*KnowledgeApplicationService, error) {
 	}
 
 	var n2s nl2sql.NL2SQL
-	if n2sChatModel, _, err := getBuiltinChatModel(ctx, "NL2SQL_"); err != nil {
+	if n2sChatModel, _, err := internal.GetBuiltinChatModel(ctx, "NL2SQL_"); err != nil {
 		return nil, err
 	} else {
 		filePath := filepath.Join(root, "resources/conf/prompt/nl2sql_template_jinja2.json")
@@ -160,7 +153,7 @@ func InitService(c *ServiceComponents) (*KnowledgeApplicationService, error) {
 		}
 	}
 
-	imageAnnoChatModel, configured, err := getBuiltinChatModel(ctx, "IA_")
+	imageAnnoChatModel, configured, err := internal.GetBuiltinChatModel(ctx, "IA_")
 	if err != nil {
 		return nil, err
 	}
@@ -372,7 +365,6 @@ func getEmbedding(ctx context.Context) (embedding.Embedder, error) {
 			return nil, fmt.Errorf("init ollama embedding failed, err=%w", err)
 		}
 
-
 	case "http":
 		var (
 			httpEmbeddingBaseURL = os.Getenv("HTTP_EMBEDDING_ADDR")
@@ -394,81 +386,6 @@ func getEmbedding(ctx context.Context) (embedding.Embedder, error) {
 	return emb, nil
 }
 
-func getBuiltinChatModel(ctx context.Context, envPrefix string) (bcm chatmodel.BaseChatModel, configured bool, err error) {
-	getEnv := func(key string) string {
-		if val := os.Getenv(envPrefix + key); val != "" {
-			return val
-		}
-		return os.Getenv(key)
-	}
-
-	switch getEnv("BUILTIN_CM_TYPE") {
-	case "openai":
-		byAzure, _ := strconv.ParseBool(getEnv("BUILTIN_CM_OPENAI_BY_AZURE"))
-		bcm, err = mo.NewChatModel(ctx, &mo.ChatModelConfig{
-			APIKey:  getEnv("BUILTIN_CM_OPENAI_API_KEY"),
-			ByAzure: byAzure,
-			BaseURL: getEnv("BUILTIN_CM_OPENAI_BASE_URL"),
-			Model:   getEnv("BUILTIN_CM_OPENAI_MODEL"),
-		})
-	case "ark":
-		bcm, err = ao.NewChatModel(ctx, &ao.ChatModelConfig{
-			APIKey:  getEnv("BUILTIN_CM_ARK_API_KEY"),
-			Model:   getEnv("BUILTIN_CM_ARK_MODEL"),
-			BaseURL: getEnv("BUILTIN_CM_ARK_BASE_URL"),
-		})
-	case "deepseek":
-		bcm, err = deepseek.NewChatModel(ctx, &deepseek.ChatModelConfig{
-			APIKey:  getEnv("BUILTIN_CM_DEEPSEEK_API_KEY"),
-			BaseURL: getEnv("BUILTIN_CM_DEEPSEEK_BASE_URL"),
-			Model:   getEnv("BUILTIN_CM_DEEPSEEK_MODEL"),
-		})
-	case "ollama":
-		bcm, err = ollama.NewChatModel(ctx, &ollama.ChatModelConfig{
-			BaseURL: getEnv("BUILTIN_CM_OLLAMA_BASE_URL"),
-			Model:   getEnv("BUILTIN_CM_OLLAMA_MODEL"),
-		})
-	case "qwen":
-		bcm, err = qwen.NewChatModel(ctx, &qwen.ChatModelConfig{
-			APIKey:  getEnv("BUILTIN_CM_QWEN_API_KEY"),
-			BaseURL: getEnv("BUILTIN_CM_QWEN_BASE_URL"),
-			Model:   getEnv("BUILTIN_CM_QWEN_MODEL"),
-		})
-	case "gemini":
-		backend, convErr := strconv.ParseInt(getEnv("BUILTIN_CM_GEMINI_BACKEND"), 10, 64)
-		if convErr != nil {
-			return nil, false, convErr
-		}
-		c, clientErr := genai.NewClient(ctx, &genai.ClientConfig{
-			APIKey:   getEnv("BUILTIN_CM_GEMINI_API_KEY"),
-			Backend:  genai.Backend(backend),
-			Project:  getEnv("BUILTIN_CM_GEMINI_PROJECT"),
-			Location: getEnv("BUILTIN_CM_GEMINI_LOCATION"),
-			HTTPOptions: genai.HTTPOptions{
-				BaseURL: getEnv("BUILTIN_CM_GEMINI_BASE_URL"),
-			},
-		})
-		if clientErr != nil {
-			return nil, false, clientErr
-		}
-		bcm, err = gemini.NewChatModel(ctx, &gemini.Config{
-			Client: c,
-			Model:  getEnv("BUILTIN_CM_GEMINI_MODEL"),
-		})
-	default:
-		// accept builtin chat model not configured
-	}
-
-	if err != nil {
-		return nil, false, fmt.Errorf("knowledge init openai chat mode failed, %w", err)
-	}
-	if bcm != nil {
-		configured = true
-	}
-
-	return
-}
-
 func readJinja2PromptTemplate(jsonFilePath string) (prompt.ChatTemplate, error) {
 	b, err := os.ReadFile(jsonFilePath)
 	if err != nil {

backend/application/workflow/init.go

@@ -17,7 +17,11 @@
 package workflow
 
 import (
+	"context"
+
 	"github.com/cloudwego/eino/compose"
+	"github.com/coze-dev/coze-studio/backend/application/internal"
+	"github.com/coze-dev/coze-studio/backend/pkg/logs"
 	"github.com/redis/go-redis/v9"
 	"gorm.io/gorm"
 
@@ -64,9 +68,16 @@ type ServiceComponents struct {
 	CodeRunner         coderunner.Runner
 }
 
-func InitService(components *ServiceComponents) *ApplicationService {
+func InitService(ctx context.Context, components *ServiceComponents) (*ApplicationService, error) {
+	bcm, ok, err := internal.GetBuiltinChatModel(ctx, "WKR_")
+	if err != nil {
+		return nil, err
+	}
+	if !ok {
+		logs.CtxWarnf(ctx, "workflow builtin chat model for knowledge recall not configured")
+	}
 	workflowRepo := service.NewWorkflowRepository(components.IDGen, components.DB, components.Cache,
-		components.Tos, components.CPStore)
+		components.Tos, components.CPStore, bcm)
 	workflow.SetRepository(workflowRepo)
 
 	workflowDomainSVC := service.NewWorkflowService(workflowRepo)
@@ -84,5 +95,5 @@ func InitService(components *ServiceComponents) *ApplicationService {
 	SVC.TosClient = components.Tos
 	SVC.IDGenerator = components.IDGen
 
-	return SVC
+	return SVC, err
 }

backend/domain/workflow/interface.go

@@ -19,6 +19,7 @@ package workflow
 import (
 	"context"
 
+	"github.com/cloudwego/eino/components/model"
 	"github.com/cloudwego/eino/compose"
 
 	"github.com/coze-dev/coze-studio/backend/api/model/ocean/cloud/workflow"
@@ -96,6 +97,8 @@ type Repository interface {
 
 	compose.CheckPointStore
 	idgen.IDGenerator
+
+	GetKnowledgeRecallChatModel() model.BaseChatModel
 }
 
 var repositorySingleton Repository

backend/domain/workflow/internal/compose/test/question_answer_test.go

@@ -102,7 +102,7 @@ func TestQuestionAnswer(t *testing.T) {
 		mockTos := storageMock.NewMockStorage(ctrl)
 		mockTos.EXPECT().GetObjectUrl(gomock.Any(), gomock.Any(), gomock.Any()).Return("", nil).AnyTimes()
 		repo := repo2.NewRepository(mockIDGen, db, redisClient, mockTos,
-			checkpoint.NewRedisStore(redisClient))
+			checkpoint.NewRedisStore(redisClient), nil)
 		mockey.Mock(workflow.GetRepository).Return(repo).Build()
 
 		t.Run("answer directly, no structured output", func(t *testing.T) {

backend/domain/workflow/internal/compose/to_node.go

@@ -223,9 +223,9 @@ func (s *NodeSchema) ToLLMConfig(ctx context.Context) (*llm.Config, error) {
 		}
 
 		if fcParams.KnowledgeFCParam != nil && len(fcParams.KnowledgeFCParam.KnowledgeList) > 0 {
-			kwChatModel, err := knowledgeRecallChatModel(ctx)
+			kwChatModel := workflow2.GetRepository().GetKnowledgeRecallChatModel()
-			if err != nil {
+			if kwChatModel == nil {
-				return nil, err
+				return nil, fmt.Errorf("workflow builtin chat model for knowledge recall not configured")
 			}
 			knowledgeOperator := crossknowledge.GetKnowledgeOperator()
 			setting := fcParams.KnowledgeFCParam.GlobalSetting
@@ -650,15 +650,3 @@ func totRetrievalSearchType(s int64) (crossknowledge.SearchType, error) {
 		return "", fmt.Errorf("invalid retrieval search type %v", s)
 	}
 }
-
-// knowledgeRecallChatModel the chat model used by the knowledge base recall in the LLM node, not the user-configured model
-func knowledgeRecallChatModel(ctx context.Context) (einomodel.BaseChatModel, error) {
-	defaultChatModelParma := &model.LLMParams{
-		ModelName:   "豆包·1.5·Pro·32k",
-		ModelType:   1,
-		Temperature: ptr.Of(0.5),
-		MaxTokens:   4096,
-	}
-	m, _, err := model.GetManager().GetModel(ctx, defaultChatModelParma)
-	return m, err
-}

backend/domain/workflow/internal/repo/repository.go

@@ -41,6 +41,7 @@ import (
 	"github.com/coze-dev/coze-studio/backend/domain/workflow/internal/execute"
 	"github.com/coze-dev/coze-studio/backend/domain/workflow/internal/repo/dal/model"
 	"github.com/coze-dev/coze-studio/backend/domain/workflow/internal/repo/dal/query"
+	cm "github.com/coze-dev/coze-studio/backend/infra/contract/chatmodel"
 	"github.com/coze-dev/coze-studio/backend/infra/contract/idgen"
 	"github.com/coze-dev/coze-studio/backend/infra/contract/storage"
 	"github.com/coze-dev/coze-studio/backend/pkg/errorx"
@@ -66,10 +67,11 @@ type RepositoryImpl struct {
 	workflow.InterruptEventStore
 	workflow.CancelSignalStore
 	workflow.ExecuteHistoryStore
+	builtinModel cm.BaseChatModel
 }
 
 func NewRepository(idgen idgen.IDGenerator, db *gorm.DB, redis *redis.Client, tos storage.Storage,
-	cpStore einoCompose.CheckPointStore) workflow.Repository {
+	cpStore einoCompose.CheckPointStore, chatModel cm.BaseChatModel) workflow.Repository {
 	return &RepositoryImpl{
 		IDGenerator:     idgen,
 		query:           query.Use(db),
@@ -86,6 +88,7 @@ func NewRepository(idgen idgen.IDGenerator, db *gorm.DB, redis *redis.Client, to
 			query: query.Use(db),
 			redis: redis,
 		},
+		builtinModel: chatModel,
 	}
 }
 
@@ -1583,6 +1586,10 @@ func (r *RepositoryImpl) BatchCreateConnectorWorkflowVersion(ctx context.Context
 	return nil
 }
 
+func (r *RepositoryImpl) GetKnowledgeRecallChatModel() cm.BaseChatModel {
+	return r.builtinModel
+}
+
 func filterDisabledAPIParameters(parametersCfg []*workflow3.APIParameter, m map[string]any) map[string]any {
 	result := make(map[string]any, len(m))
 	responseParameterMap := slices.ToMap(parametersCfg, func(p *workflow3.APIParameter) (string, *workflow3.APIParameter) {

backend/domain/workflow/service/service_impl.go

@@ -39,6 +39,7 @@ import (
 	"github.com/coze-dev/coze-studio/backend/domain/workflow/internal/canvas/adaptor"
 	"github.com/coze-dev/coze-studio/backend/domain/workflow/internal/compose"
 	"github.com/coze-dev/coze-studio/backend/domain/workflow/internal/repo"
+	"github.com/coze-dev/coze-studio/backend/infra/contract/chatmodel"
 	"github.com/coze-dev/coze-studio/backend/infra/contract/idgen"
 	"github.com/coze-dev/coze-studio/backend/infra/contract/storage"
 	"github.com/coze-dev/coze-studio/backend/pkg/lang/ptr"
@@ -67,8 +68,8 @@ func NewWorkflowService(repo workflow.Repository) workflow.Service {
 }
 
 func NewWorkflowRepository(idgen idgen.IDGenerator, db *gorm.DB, redis *redis.Client, tos storage.Storage,
-	cpStore einoCompose.CheckPointStore) workflow.Repository {
+	cpStore einoCompose.CheckPointStore, chatModel chatmodel.BaseChatModel) workflow.Repository {
-	return repo.NewRepository(idgen, db, redis, tos, cpStore)
+	return repo.NewRepository(idgen, db, redis, tos, cpStore, chatModel)
 }
 
 func (i *impl) ListNodeMeta(ctx context.Context, nodeTypes map[entity.NodeType]bool) (map[string][]*entity.NodeTypeMeta, []entity.Category, error) {

backend/internal/mock/domain/workflow/interface.go

@@ -13,6 +13,7 @@ import (
 	context "context"
 	reflect "reflect"
 
+	model "github.com/cloudwego/eino/components/model"
 	compose "github.com/cloudwego/eino/compose"
 	schema "github.com/cloudwego/eino/schema"
 	workflow "github.com/coze-dev/coze-studio/backend/api/model/ocean/cloud/workflow"
@@ -26,7 +27,6 @@ import (
 type MockService struct {
 	ctrl     *gomock.Controller
 	recorder *MockServiceMockRecorder
-	isgomock struct{}
 }
 
 // MockServiceMockRecorder is the mock recorder for MockService.
@@ -364,17 +364,17 @@ func (mr *MockServiceMockRecorder) ReleaseApplicationWorkflows(ctx, appID, confi
 }
 
 // Save mocks base method.
-func (m *MockService) Save(ctx context.Context, id int64, arg2 string) error {
+func (m *MockService) Save(ctx context.Context, id int64, schema string) error {
 	m.ctrl.T.Helper()
-	ret := m.ctrl.Call(m, "Save", ctx, id, arg2)
+	ret := m.ctrl.Call(m, "Save", ctx, id, schema)
 	ret0, _ := ret[0].(error)
 	return ret0
 }
 
 // Save indicates an expected call of Save.
-func (mr *MockServiceMockRecorder) Save(ctx, id, arg2 any) *gomock.Call {
+func (mr *MockServiceMockRecorder) Save(ctx, id, schema any) *gomock.Call {
 	mr.mock.ctrl.T.Helper()
-	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "Save", reflect.TypeOf((*MockService)(nil).Save), ctx, id, arg2)
+	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "Save", reflect.TypeOf((*MockService)(nil).Save), ctx, id, schema)
 }
 
 // StreamExecute mocks base method.
@@ -528,7 +528,6 @@ func (mr *MockServiceMockRecorder) WorkflowAsModelTool(ctx, policies any) *gomoc
 type MockRepository struct {
 	ctrl     *gomock.Controller
 	recorder *MockRepositoryMockRecorder
-	isgomock struct{}
 }
 
 // MockRepositoryMockRecorder is the mock recorder for MockRepository.
@@ -548,6 +547,20 @@ func (m *MockRepository) EXPECT() *MockRepositoryMockRecorder {
 	return m.recorder
 }
 
+// BatchCreateConnectorWorkflowVersion mocks base method.
+func (m *MockRepository) BatchCreateConnectorWorkflowVersion(ctx context.Context, appID, connectorID int64, workflowIDs []int64, version string) error {
+	m.ctrl.T.Helper()
+	ret := m.ctrl.Call(m, "BatchCreateConnectorWorkflowVersion", ctx, appID, connectorID, workflowIDs, version)
+	ret0, _ := ret[0].(error)
+	return ret0
+}
+
+// BatchCreateConnectorWorkflowVersion indicates an expected call of BatchCreateConnectorWorkflowVersion.
+func (mr *MockRepositoryMockRecorder) BatchCreateConnectorWorkflowVersion(ctx, appID, connectorID, workflowIDs, version any) *gomock.Call {
+	mr.mock.ctrl.T.Helper()
+	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "BatchCreateConnectorWorkflowVersion", reflect.TypeOf((*MockRepository)(nil).BatchCreateConnectorWorkflowVersion), ctx, appID, connectorID, workflowIDs, version)
+}
+
 // CancelAllRunningNodes mocks base method.
 func (m *MockRepository) CancelAllRunningNodes(ctx context.Context, wfExeID int64) error {
 	m.ctrl.T.Helper()
@@ -784,6 +797,20 @@ func (mr *MockRepositoryMockRecorder) GetFirstInterruptEvent(ctx, wfExeID any) *
 	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "GetFirstInterruptEvent", reflect.TypeOf((*MockRepository)(nil).GetFirstInterruptEvent), ctx, wfExeID)
 }
 
+// GetKnowledgeRecallChatModel mocks base method.
+func (m *MockRepository) GetKnowledgeRecallChatModel() model.BaseChatModel {
+	m.ctrl.T.Helper()
+	ret := m.ctrl.Call(m, "GetKnowledgeRecallChatModel")
+	ret0, _ := ret[0].(model.BaseChatModel)
+	return ret0
+}
+
+// GetKnowledgeRecallChatModel indicates an expected call of GetKnowledgeRecallChatModel.
+func (mr *MockRepositoryMockRecorder) GetKnowledgeRecallChatModel() *gomock.Call {
+	mr.mock.ctrl.T.Helper()
+	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "GetKnowledgeRecallChatModel", reflect.TypeOf((*MockRepository)(nil).GetKnowledgeRecallChatModel))
+}
+
 // GetLatestVersion mocks base method.
 func (m *MockRepository) GetLatestVersion(ctx context.Context, id int64) (*vo.VersionInfo, error) {
 	m.ctrl.T.Helper()
@@ -936,6 +963,21 @@ func (mr *MockRepositoryMockRecorder) GetWorkflowExecution(ctx, id any) *gomock.
 	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "GetWorkflowExecution", reflect.TypeOf((*MockRepository)(nil).GetWorkflowExecution), ctx, id)
 }
 
+// IsApplicationConnectorWorkflowVersion mocks base method.
+func (m *MockRepository) IsApplicationConnectorWorkflowVersion(ctx context.Context, connectorID, workflowID int64, version string) (bool, error) {
+	m.ctrl.T.Helper()
+	ret := m.ctrl.Call(m, "IsApplicationConnectorWorkflowVersion", ctx, connectorID, workflowID, version)
+	ret0, _ := ret[0].(bool)
+	ret1, _ := ret[1].(error)
+	return ret0, ret1
+}
+
+// IsApplicationConnectorWorkflowVersion indicates an expected call of IsApplicationConnectorWorkflowVersion.
+func (mr *MockRepositoryMockRecorder) IsApplicationConnectorWorkflowVersion(ctx, connectorID, workflowID, version any) *gomock.Call {
+	mr.mock.ctrl.T.Helper()
+	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "IsApplicationConnectorWorkflowVersion", reflect.TypeOf((*MockRepository)(nil).IsApplicationConnectorWorkflowVersion), ctx, connectorID, workflowID, version)
+}
+
 // ListInterruptEvents mocks base method.
 func (m *MockRepository) ListInterruptEvents(ctx context.Context, wfExeID int64) ([]*entity.InterruptEvent, error) {
 	m.ctrl.T.Helper()
@@ -1172,6 +1214,20 @@ func (mr *MockRepositoryMockRecorder) UpdateNodeExecution(ctx, execution any) *g
 	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "UpdateNodeExecution", reflect.TypeOf((*MockRepository)(nil).UpdateNodeExecution), ctx, execution)
 }
 
+// UpdateNodeExecutionStreaming mocks base method.
+func (m *MockRepository) UpdateNodeExecutionStreaming(ctx context.Context, execution *entity.NodeExecution) error {
+	m.ctrl.T.Helper()
+	ret := m.ctrl.Call(m, "UpdateNodeExecutionStreaming", ctx, execution)
+	ret0, _ := ret[0].(error)
+	return ret0
+}
+
+// UpdateNodeExecutionStreaming indicates an expected call of UpdateNodeExecutionStreaming.
+func (mr *MockRepositoryMockRecorder) UpdateNodeExecutionStreaming(ctx, execution any) *gomock.Call {
+	mr.mock.ctrl.T.Helper()
+	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "UpdateNodeExecutionStreaming", reflect.TypeOf((*MockRepository)(nil).UpdateNodeExecutionStreaming), ctx, execution)
+}
+
 // UpdateWorkflowDraftTestRunSuccess mocks base method.
 func (m *MockRepository) UpdateWorkflowDraftTestRunSuccess(ctx context.Context, id int64) error {
 	m.ctrl.T.Helper()

docker/.env.example

@@ -101,7 +101,7 @@ export EMBEDDING_TYPE="ark"
 export OPENAI_EMBEDDING_BASE_URL=""    # (string) OpenAI base_url
 export OPENAI_EMBEDDING_MODEL=""       # (string) OpenAI embedding model
 export OPENAI_EMBEDDING_API_KEY=""     # (string) OpenAI api_key
-export OPENAI_EMBEDDING_BY_AZURE=true  # (bool) OpenAI by_azure
+export OPENAI_EMBEDDING_BY_AZURE=false  # (bool) OpenAI by_azure
 export OPENAI_EMBEDDING_API_VERSION="" # OpenAI azure api version
 export OPENAI_EMBEDDING_DIMS=1024      # (int) 向量维度
 export OPENAI_EMBEDDING_REQUEST_DIMS=1024
@@ -140,17 +140,18 @@ export MODEL_ID_0=""                # model name for connection
 export MODEL_API_KEY_0=""           # model api key
 export MODEL_BASE_URL_0=""           # model base url
 
-# Model for knowledge nl2sql, messages2query (rewrite), image annotation
+# Model for knowledge nl2sql, messages2query (rewrite), image annotation, workflow knowledge recall
 # add prefix to assign specific model, downgrade to default config when prefix is not configured:
-# 1. nl2sql:            NL2SQL_ (e.g. NL2SQL_BUILTIN_CM_TYPE)
+# 1. nl2sql:                    NL2SQL_ (e.g. NL2SQL_BUILTIN_CM_TYPE)
-# 2. messages2query:    M2Q_    (e.g. M2Q_BUILTIN_CM_TYPE)
+# 2. messages2query:            M2Q_    (e.g. M2Q_BUILTIN_CM_TYPE)
-# 3. image annotation:  IA_     (e.g. IA_BUILTIN_CM_TYPE)
+# 3. image annotation:          IA_     (e.g. IA_BUILTIN_CM_TYPE)
+# 4. workflow knowledge recall: WKR_    (e.g. WKR_BUILTIN_CM_TYPE)
 # supported chat model type: openai / ark / deepseek / ollama / qwen / gemini
 export BUILTIN_CM_TYPE="ark"
 # type openai
 export BUILTIN_CM_OPENAI_BASE_URL=""
 export BUILTIN_CM_OPENAI_API_KEY=""
-export BUILTIN_CM_OPENAI_BY_AZURE=true
+export BUILTIN_CM_OPENAI_BY_AZURE=false
 export BUILTIN_CM_OPENAI_MODEL=""
 
 # type ark