feat: mcp支持创建清洗任务

backend/services/data-cleaning-service/src/main/java/com/datamate/cleaning/interfaces/dto/CreateCleaningTaskRequest.java

@@ -6,6 +6,7 @@
 
 import lombok.Getter;
 import lombok.Setter;
+import org.springaicommunity.mcp.annotation.McpToolParam;
 
 /**
  * CreateCleaningTaskRequest
@@ -14,21 +15,29 @@
 @Getter
 @Setter
 public class CreateCleaningTaskRequest {
+    @McpToolParam(description = "新建清洗任务名称")
+    private String name;
 
-  private String name;
+    @McpToolParam(description = "新建清洗任务描述")
+    private String description;
 
-  private String description;
+    @McpToolParam(description = "清洗任务使用的源数据集ID")
+    private String srcDatasetId;
 
-  private String srcDatasetId;
+    @McpToolParam(description = "清洗任务使用的源数据集名称")
+    private String srcDatasetName;
 
-  private String srcDatasetName;
+    @McpToolParam(description = "清洗任务创建的目标数据集名称")
+    private String destDatasetName;
 
-  private String destDatasetName;
+    @McpToolParam(description = "清洗任务创建的目标数据集类型，取值范围为TEXT/IMAGE/VIDEO/AUDIO/OTHER")
+    private String destDatasetType;
 
-  private String destDatasetType;
+    @McpToolParam(description = "清洗任务使用的模板名称，与instance参数二选一，至少指定一个，优先级更高", required = false)
+    private String templateId;
 
-  private String templateId;
-
-  private List<OperatorInstanceDto> instance = new ArrayList<>();
+    @McpToolParam(description = "清洗任务使用的算子列表，与templateId参数二选一，至少指定一个。" +
+            "注意：单个任务只能使用一种归属的算子，无法混合使用，如全部使用DataMate算子或全部使用DataJuicer算子。", required = false)
+    private List<OperatorInstanceDto> instance = new ArrayList<>();
 }
 

backend/services/data-cleaning-service/src/main/java/com/datamate/cleaning/interfaces/dto/OperatorInstanceDto.java

@@ -7,6 +7,7 @@
 
 import lombok.Getter;
 import lombok.Setter;
+import org.springaicommunity.mcp.annotation.McpToolParam;
 
 /**
  * OperatorInstance
@@ -15,17 +16,22 @@
 @Getter
 @Setter
 public class OperatorInstanceDto {
-
+    @McpToolParam(description = "算子ID")
     private String id;
 
+    @McpToolParam(description = "算子名称")
     private String name;
 
+    @McpToolParam(description = "算子输入类型，取值范围为text/image/audio/video/multimodal")
     private String inputs;
 
+    @McpToolParam(description = "算子输出类型，取值范围为text/image/audio/video/multimodal")
     private String outputs;
 
+    @McpToolParam(description = "算子所属分类的所有ID组成的列表。", required = false)
     private List<String> categories;
 
+    @McpToolParam(description = "算子需要覆盖的参数", required = false)
     private Map<String, Object> overrides = new HashMap<>();
 }
 

backend/services/data-cleaning-service/src/main/java/com/datamate/cleaning/interfaces/rest/CleaningTaskController.java

@@ -5,6 +5,8 @@
 import com.datamate.common.interfaces.PagedResponse;
 import lombok.RequiredArgsConstructor;
 import org.apache.commons.lang3.StringUtils;
+import org.springaicommunity.mcp.annotation.McpTool;
+import org.springaicommunity.mcp.annotation.McpToolParam;
 import org.springframework.web.bind.annotation.*;
 
 import java.util.List;
@@ -28,7 +30,9 @@ public PagedResponse<CleaningTaskDto> cleaningTasksGet(
     }
 
     @PostMapping
-    public CleaningTaskDto cleaningTasksPost(@RequestBody CreateCleaningTaskRequest request) {
+    @McpTool(name = "create_cleaning_task", description = "根据模板ID或算子列表创建清洗任务。")
+    public CleaningTaskDto cleaningTasksPost(@McpToolParam(description = "创建任务请求体，需要将参数包装在request对象中。")
+                                             @RequestBody CreateCleaningTaskRequest request) {
         if (request.getInstance().isEmpty() && StringUtils.isNotBlank(request.getTemplateId())) {
             request.setInstance(cleaningTaskService.getInstanceByTemplateId(request.getTemplateId()));
         }

backend/services/data-cleaning-service/src/main/java/com/datamate/cleaning/interfaces/rest/CleaningTemplateController.java

@@ -6,6 +6,8 @@
 import com.datamate.cleaning.interfaces.dto.UpdateCleaningTemplateRequest;
 import com.datamate.common.interfaces.PagedResponse;
 import lombok.RequiredArgsConstructor;
+import org.springaicommunity.mcp.annotation.McpTool;
+import org.springaicommunity.mcp.annotation.McpToolParam;
 import org.springframework.web.bind.annotation.DeleteMapping;
 import org.springframework.web.bind.annotation.GetMapping;
 import org.springframework.web.bind.annotation.PathVariable;
@@ -27,10 +29,14 @@ public class CleaningTemplateController {
     private final CleaningTemplateService cleaningTemplateService;
 
     @GetMapping
+    @McpTool(name = "query_cleaning_template", description = "查询模板列表")
     public PagedResponse<CleaningTemplateDto> cleaningTemplatesGet(
-            @RequestParam(value = "page", required = false) Integer page,
-            @RequestParam(value = "size", required = false) Integer size,
-            @RequestParam(value = "keyword", required = false) String keyword) {
+            @RequestParam(value = "page", required = false)
+            @McpToolParam(description = "页码，从0开始", required = false) Integer page,
+            @RequestParam(value = "size", required = false)
+            @McpToolParam(description = "每页大小", required = false) Integer size,
+            @RequestParam(value = "keyword", required = false)
+            @McpToolParam(description = "关键词，从名称与描述中查询", required = false) String keyword) {
         List<CleaningTemplateDto> templates = cleaningTemplateService.getTemplates(keyword);
         if (page == null || size == null) {
             return PagedResponse.of(templates.stream()

backend/services/data-management-service/src/main/java/com/datamate/datamanagement/interfaces/dto/CreateDatasetRequest.java

@@ -8,6 +8,7 @@
 import lombok.Getter;
 import lombok.NoArgsConstructor;
 import lombok.Setter;
+import org.springaicommunity.mcp.annotation.McpToolParam;
 
 import java.util.List;
 
@@ -22,19 +23,26 @@ public class CreateDatasetRequest {
     /** 数据集名称 */
     @Size(min = 1, max = 100)
     @NotBlank(message = "数据集名称不能为空")
+    @McpToolParam(description = "数据集名称")
     private String name;
     /** 数据集描述 */
     @Size(max = 500)
+    @McpToolParam(description = "数据集描述", required = false)
     private String description;
     /** 数据集类型 */
     @NotNull(message = "数据集类型不能为空")
+    @McpToolParam(description = "数据集类型，取值范围为TEXT/IMAGE/VIDEO/AUDIO/OTHER")
     private DatasetType datasetType;
     /** 标签列表 */
+    @McpToolParam(description = "标签列表", required = false)
     private List<String> tags;
     /** 数据源 */
+    @McpToolParam(description = "数据源", required = false)
     private String dataSource;
     /** 保留天数 */
+    @McpToolParam(description = "保留天数", required = false)
     private Integer retentionDays;
     /** 数据集状态 */
+    @McpToolParam(description = "数据集状态", required = false)
     private String status;
 }

backend/services/data-management-service/src/main/java/com/datamate/datamanagement/interfaces/dto/DatasetPagingQuery.java

@@ -7,6 +7,7 @@
 import lombok.Getter;
 import lombok.NoArgsConstructor;
 import lombok.Setter;
+import org.springaicommunity.mcp.annotation.McpToolParam;
 
 import java.util.ArrayList;
 import java.util.List;
@@ -23,20 +24,24 @@ public class DatasetPagingQuery extends PagingQuery {
     /**
      * 数据集类型过滤
      */
+    @McpToolParam(description = "数据集类型过滤，取值范围为text/image/video/audio/other", required = false)
     private DatasetType type;
 
     /**
      * 标签名过滤
      */
+    @McpToolParam(description = "标签名过滤", required = false)
     private List<String> tags = new ArrayList<>();
 
     /**
      * 关键词搜索（名称或描述）
      */
+    @McpToolParam(description = "关键词搜索（名称或描述）", required = false)
     private String keyword;
 
     /**
      * 状态过滤
      */
+    @McpToolParam(description = "状态过滤", required = false)
     private DatasetStatusType status;
 }

backend/services/data-management-service/src/main/java/com/datamate/datamanagement/interfaces/rest/DatasetController.java

@@ -11,6 +11,7 @@
 import jakarta.validation.Valid;
 import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
+import org.springaicommunity.mcp.annotation.McpTool;
 import org.springframework.http.HttpStatus;
 import org.springframework.http.ResponseEntity;
 import org.springframework.web.bind.annotation.*;
@@ -34,6 +35,7 @@ public class DatasetController {
      * @return 分页的数据集列表
      */
     @GetMapping
+    @McpTool(name = "query_datasets", description = "根据参数查询满足条件的数据集列表")
     public PagedResponse<DatasetResponse> getDatasets(DatasetPagingQuery query) {
         return datasetApplicationService.getDatasets(query);
     }
@@ -45,6 +47,7 @@ public PagedResponse<DatasetResponse> getDatasets(DatasetPagingQuery query) {
      * @return 创建的数据集响应
      */
     @PostMapping
+    @McpTool(name = "create_dataset", description = "创建数据集")
     public DatasetResponse createDataset(@RequestBody @Valid CreateDatasetRequest createDatasetRequest) {
         Dataset dataset = datasetApplicationService.createDataset(createDatasetRequest);
         return DatasetConverter.INSTANCE.convertToResponse(dataset);

backend/services/main-application/src/main/resources/application.yml

@@ -67,6 +67,19 @@ spring:
           min-idle: 5
           max-wait: 1000ms
 
+  ai:
+    mcp:
+      server:
+        enabled: true
+        name: datamate-mcp-server
+        protocol: STATELESS
+        capabilities:
+          completion: false
+          resource: false
+          prompt: false
+          tool: true
+        mcp-endpoint: /mcp
+
 
 # MyBatis配置（需在顶层，不在 spring 下）
 mybatis-plus:

backend/services/operator-market-service/src/main/java/com/datamate/operator/interfaces/dto/OperatorsListPostRequest.java

@@ -7,6 +7,7 @@
 import com.datamate.common.interfaces.PagingQuery;
 import lombok.Getter;
 import lombok.Setter;
+import org.springaicommunity.mcp.annotation.McpToolParam;
 
 /**
  * OperatorsListPostRequest
@@ -15,12 +16,16 @@
 @Getter
 @Setter
 public class OperatorsListPostRequest extends PagingQuery {
-  private List<List<String>> categories = new ArrayList<>();
+    @McpToolParam(description = "算子分类id列表，每个父分类下的id放到一个列表中，最后汇总成一个大的列表", required = false)
+    private List<List<String>> categories = new ArrayList<>();
 
-  private String keyword;
+    @McpToolParam(description = "算子关键词，支持查询算子名称和算子描述关键词查询", required = false)
+    private String keyword;
 
-  private String labelName;
+    @McpToolParam(description = "算子关联的标签名称，当前暂不支持", required = false)
+    private String labelName;
 
-  private Boolean isStar;
+    @McpToolParam(description = "算子是否被收藏", required = false)
+    private Boolean isStar;
 }
 

backend/services/operator-market-service/src/main/java/com/datamate/operator/interfaces/rest/CategoryController.java

@@ -6,6 +6,7 @@
 import com.datamate.operator.interfaces.dto.CategoryTreePagedResponse;
 import com.datamate.operator.interfaces.dto.CategoryTreeResponse;
 import lombok.RequiredArgsConstructor;
+import org.springaicommunity.mcp.annotation.McpTool;
 import org.springframework.web.bind.annotation.GetMapping;
 import org.springframework.web.bind.annotation.RequestMapping;
 import org.springframework.web.bind.annotation.RestController;
@@ -22,6 +23,8 @@ public class CategoryController {
     private final OperatorRepository operatorRepo;
 
     @GetMapping("/tree")
+    @McpTool(name = "query_category_tree",
+            description = "算子树状分类查询，获取包含分组维度（如语言、模态）及资源统计数量的分页层级分类数据。")
     public PagedResponse<CategoryTreeResponse> categoryTreeGet() {
         List<CategoryTreeResponse> allCategories = categoryService.getAllCategories();
         return CategoryTreePagedResponse.of(allCategories, operatorRepo.countOperatorByStar(true));

backend/services/operator-market-service/src/main/java/com/datamate/operator/interfaces/rest/OperatorController.java

@@ -7,6 +7,7 @@
 import com.datamate.operator.interfaces.dto.OperatorsListPostRequest;
 import com.datamate.operator.interfaces.dto.UploadOperatorRequest;
 import lombok.RequiredArgsConstructor;
+import org.springaicommunity.mcp.annotation.McpTool;
 import org.springframework.core.io.Resource;
 import org.springframework.http.HttpHeaders;
 import org.springframework.http.HttpStatus;
@@ -24,6 +25,7 @@ public class OperatorController {
     private final OperatorService operatorService;
 
     @PostMapping("/list")
+    @McpTool(name = "query_operator_list", description = "根据参数查询算子列表")
     public PagedResponse<OperatorDto> operatorsListPost(@RequestBody OperatorsListPostRequest request) {
         List<List<String>> categories = request.getCategories();
         List<OperatorDto> responses = operatorService.getOperators(request.getPage(), request.getSize(),

backend/services/pom.xml

@@ -87,6 +87,12 @@
             <groupId>org.apache.poi</groupId>
             <artifactId>poi</artifactId>
         </dependency>
+
+        <dependency>
+            <groupId>org.springframework.ai</groupId>
+            <artifactId>spring-ai-starter-mcp-server-webmvc</artifactId>
+            <version>1.1.2</version>
+        </dependency>
     </dependencies>
 
     <build>

backend/shared/domain-common/src/main/java/com/datamate/common/interfaces/PagingQuery.java

@@ -2,18 +2,21 @@
 
 import lombok.Getter;
 import lombok.NoArgsConstructor;
+import org.springaicommunity.mcp.annotation.McpToolParam;
 
 @Getter
 @NoArgsConstructor
 public class PagingQuery {
     /**
      * 页码，从0开始
      */
+    @McpToolParam(description = "页码，从0开始")
     private Integer page = 0;
 
     /**
      * 每页大小
      */
+    @McpToolParam(description = "每页大小")
     private Integer size = 20;
 
     public void setPage(Integer page) {

runtime/datamate-python/app/main.py

@@ -3,6 +3,7 @@
 
 from fastapi import FastAPI, HTTPException
 from fastapi.exceptions import RequestValidationError
+from fastapi_mcp import FastApiMCP
 from sqlalchemy import text
 from starlette.exceptions import HTTPException as StarletteHTTPException
 
@@ -102,6 +103,9 @@ async def root():
         }
     )
 
+mcp = FastApiMCP(app, name="DataMate MCP", description="DataMate python mcp server", include_tags=["mcp"])
+mcp.mount_http(mount_path="/api/mcp")
+
 if __name__ == "__main__":
     import uvicorn
 

runtime/datamate-python/app/module/collection/interface/collection.py

@@ -24,7 +24,7 @@
 logger = get_logger(__name__)
 
 
-@router.post("", response_model=StandardResponse[CollectionTaskBase])
+@router.post("", response_model=StandardResponse[CollectionTaskBase], operation_id="create_collect_task", tags=["mcp"])
 async def create_task(
     request: CollectionTaskCreate,
     db: AsyncSession = Depends(get_db)