All Products
Search
Document Center

DataWorks:DataWorks Agent for third-party clients

Last Updated:Jan 27, 2026

Built on the MCP, DataWorks Agent connects to DataWorks MCP server and other big data MCP Servers (such as Hologres MCP Server). It enables data development, task operation, and data integration capabilities in DataWorks through natural language interaction.

Important

This feature requires third-party clients. For a more integrated Agent experience, see DataWorks Agent.

Overview

DataWorks Agent supports big data development via natural language interaction in an intelligent chat window. DataWorks Agent uses LLMs to parse requirements and calls MCP Server capabilities to execute tasks.

For example, enter "How many workspaces do I have" in the DataWorks Agent chat window. The Agent uses the LLM to parse the request, calls the ListProjects tool provided by DataWorks MCP Server (which uses the built-in DataWorks OpenAPI), and returns the results. When a task is complex, the LLM will interact with the MCP Server multiple times.

DataWorks Agent integrates DataWorks MCP Server and connects to other MCP Servers. You can also choose your own LLM (such as Qwen, DeepSeek, OpenAI, etc.).

image

Use the following prompts to explore more DataWorks Agent scenarios.

Scope

Sample prompts

Data Studio

[Query tasks] Find all paused odps_sql nodes in the current workspace directory.

[Rename tasks] Rename the nodes mentioned above to: InvalidNode_ToDelete. Suffix node names with serial numbers.

[Create tasks] Create 5 odps_sql nodes in the project directory of the current workspace. The names should start with MCP_Demo, be connected by underscores, and end with an auto-incrementing number starting from 01.

Task Operation

[Rerun failed tasks] Check tasks that failed on 20250330 in the current workspace and rerun these instances. Note that bizdate is a timestamp in milliseconds.

[Query failed instances] Check instances with the status "failed" on 20250331 in the project of the current workspace. The format of bizdate is a timestamp in milliseconds.

[Analyze rerun properties] Analyze the rerun properties of these tasks to see if they can be rerun. If yes, rerun the tasks.

Data Integration

[Sync a single MySQL table to MaxCompute] Create an offline synchronization task named mcp_test_mysql to mcp_test_maxcompute in the current workspace. Use the resource group mcp_test_res. Synchronize the "users" table from the "test_db" database under the "mcp_test_mysql" data source to the table with the same name in "mcp_test_maxcompute". Use same-name mapping for field mapping. Then run this task. Report the synchronization status.

Data Analysis

[Sales analysis] Analyze the sales trend of the top 10 best-selling products in the "order" table for this month.

Limitations

  • This feature is supported only in the personal development environment.

    Important

    After the personal development environment is restarted, you need to reinstall the MCP Server function. Proceed with caution.

  • This feature is supported only in workspaces in Use Data Studio (New Version).

Billing

DataWorks Agent usage incurs OpenAPI billing and model token billing costs.

Quick start with DataWorks Agent

After configuring DataWorks Agent, click the image icon in the upper-right corner of the Cline page to open the chat interface. Example: Create ODPS SQL tasks.

image

Enter the prompt: Create 5 odps_sql nodes in the project directory of the current workspace. The names should start with MCP_Demo, be connected by underscores, and end with an auto-incrementing number starting from 01.

The Agent breaks down the steps and runs them:

  1. Identifies that the CreateNode API is needed but the workspace ID is missing, and prompts for input.

  2. Calls the CreateNode API to create 5 ODPS_SQL nodes.

  3. The CreateNode API is called successfully, returning a request ID and node IDs.

  4. Calls the ListNodes API to list nodes in the workspace to confirm successful creation.

    Note
    • The system prompts for confirmation. Click Approve to proceed or Reject to cancel.

    • The disassembly process for key steps may vary slightly with each run, and different LLMs may parse it differently. Please refer to the actual situation.

After completion, you will get 5 ODPS_SQL nodes starting with MCP_Demo.

image

Configure DataWorks Agent

DataWorks Agent uses the MCP Client extension (such as Cline) as the chat interface and connects to DataWorks MCP Server and other Alibaba Cloud MCP Servers via MCP Server configuration.

Note

You can connect to more open source MCP Servers as needed to enhance the capabilities of DataWorks Agent.

Prerequisites

Step 1: Prepare personal development environment

To enable and enter the personal development environment:

  1. Go to the Workspaces page in the DataWorks console. In the top navigation bar, select a desired region. Find the desired workspace and choose Shortcuts > Data Studio in the Actions column.

  2. Click the image icon next to Personal Development Environment in the top navigation bar to check the status of your personal development environment instance and enter the environment.

    • If the status is In operation: Click the running personal development environment to enter the Welcome Page of the personal development environment.

    • Status is other: Click Management Environment in the pop-up window. Find your instance on the Personal Development Environment Instances page, click Start in the Actions column, and wait for the instance status to change to In operation. Click the personal development environment to enter the Welcome Page of the personal development environment.

    Note

    The image icon indicates you have entered the personal development environment.

Step 2: Configure Agent

After entering the personal development environment, you can configure DataWorks Agent by following these steps. This article uses Cline as an example.

(Optional) Upgrade program engine

If you are using a previous personal development environment or have already installed the Cline extension in the environment, you need to upgrade the existing underlying engine to use the extension features. You can skip this step if you have already upgraded.

One-click upgrade: After entering the personal development environment, if you encounter a pop-up prompting for underlying engine upgrade adaptation, click One-click Upgrade to complete the adaptation upgrade.

Command upgrade: Click the image icon on the left side of the bottom toolbar to enter the terminal command bar. Enter the following upgrade commands in the command bar and press Enter to upgrade.

wget https://nodejs.org/dist/v20.19.0/node-v20.19.0-linux-x64.tar.xz
tar xf node-v20.19.0-linux-x64.tar.xz
mv /etc/dsw/node /etc/dsw/node14
mv node-v20.19.0-linux-x64 /etc/dsw/node

bash <(curl -s https://dataworks-notebook-${REGION}.oss-${REGION}.aliyuncs.com/public-datasets/aone-release/dwcode-server/scripts/update.sh)  0.2.169

After the upgrade is complete, click Reload to apply the latest changes.

Install Cline extension

To install the Cline extension:

  1. Click the image icon on the left navigation bar of the personal development environment Welcome Page to enter the Extensions page.

  2. Enter Cline in the search box on the Extensions page.

  3. Locate the Cline extension.

  4. Click Install at the bottom right of the Cline extension and wait for the installation to complete.

  5. After installing the Cline extension, you can right-click the image icon on the left navigation bar to Configure Keybindings, or select Move to > Secondary Sidebar or Move to > Panel according to your preference.

Configure LLM API key

After installing the Cline extension, follow these steps to configure the API Key information. This example uses the OpenAI Compatible mode to connect to the Alibaba Cloud Model Studio (Bailian) API.

Note
  • For other connection modes, please configure relevant parameters according to the actual interface.

  • Currently, only the Use your own API key configuration method is supported. The Get Started for Free configuration method is not supported.

  1. Click the image icon on the left navigation bar of the personal development environment Welcome Page to enter the Cline page.

  2. Click Use your own API key in the Cline page and configure it according to the parameter descriptions in the table below.

    Parameter

    Description

    API Provider

    Specify the API service provider you want to use. Select OpenAI Compatible. This connects to the Alibaba Cloud Model Studio API via an OpenAI-compatible interface.

    Base URL

    The base URL of the API service, used to specify the root address of API requests.

    For example, the API endpoint address compatible with OpenAI provided by Alibaba Cloud Model Studio: https://dashscope-intl.aliyuncs.com/compatible-mode/v1.

    API Key

    The key used for identity authentication to ensure your request is legitimate.

    Obtain the API Key from the Alibaba Cloud Model Studio console.

    Model ID

    Specify the specific model you want to use. Different models may have different functions and performance.

    Select qwen-coder-plus or qwen-plus. These two options represent different pre-trained models:

    • qwen-coder-plus: Suitable for code generation and programming tasks.

    • qwen-plus: Suitable for general text generation and processing tasks.

  3. Click the Let's go! button below to complete the API Key configuration.

Access MCP Server

After configuring the API Key, you can follow these steps to complete the relevant configuration for the MCP Server.

  1. Click the image icon in the upper-right corner of the Cline page to enter the Marketplace tab of MCP Server.

  2. Click to switch to the Installed tab.

  3. Click Configure MCP Servers. Add the following JSON to the configuration file on the right.

    Note

    If you need to use Hologres capabilities in DataWorks Agent, please access Hologres MCP Server.

    {
      "mcpServers": {
        "alibabacloud-dataworks-mcp-server": {
          "command": "npx",
          "args": [
            "alibabacloud-dataworks-mcp-server"
          ],
          "env": {
            "REGION":"cn-beijing", 
            "ALIBABA_CLOUD_CREDENTIALS_URI":"http://localhost:7002/api/v1/credentials/0"
          },
          "disabled": false,
          "autoApprove": [],
          "timeout": 60
        }
      }
    }

    Parameter

    Description

    command

    Enter npx here, indicating the command method provided by dataworks-mcp-server.

    args

    Enter the npm package name alibabacloud-dataworks-mcp-server here, indicating the command argument for dataworks-mcp-server.

    env

    REGION

    Set this to your region ID.

    ALIBABA_CLOUD_CREDENTIALS_URI

    Specify the URI for Alibaba Cloud credentials.

    Important

    This parameter is only effective in the DataWorks personal development environment and is used to obtain Alibaba Cloud user identity authentication.

    TOOL_CATEGORIES

    Configure the whitelist for Tool categories. Please enter the OpenAPI categories here, separated by half-width commas.

    For example: "TOOL_CATEGORIES":"Data Source,Workspace Management,Resource Group Management,Data Map,Data Integration,Data Development (New),Operation Center,DataService,Open Platform,Data Quality,Label Management,Security Center,SERVER_IDE_DEFAULT".

    Note

    SERVER_IDE_DEFAULT is the default Tools on the personal development environment. Other business categories can be viewed in the left directory tree of the DataWorks - OpenAPI Overview page.

    TOOL_NAMES

    Configure the whitelist for Tool names. Please enter the OpenAPI names here, separated by half-width commas.

    For example: "TOOL_NAMES":"ListProjects,CreateNode,UpdateNode".

    Note

    TOOL_NAMES can be viewed in the DataWorks - OpenAPI Overview.

  4. After editing the configuration information, press Ctrl+S to save and load the MCP Server information you configured.

  5. The loaded tools (shown below) confirm that alibabacloud-dataworks-mcp-server has been successfully installed, and you can start using the DataWorks MCP Server function.

    Note

    If the following information fails to load, please confirm whether you have upgraded the engine.

    image

FAQ

  • When executing a preset prompt on the MCP server, if an API request hangs, it may be because your program engine version is low and has compatibility issues with the current system. To resolve this, you need to upgrade the program engine.

  • If the model response is slow, you can reduce the number of enabled MCP Servers, or add TOOL_CATEGORIES and TOOL_NAMES in the env parameter of the Access MCP Server configuration file to reduce the number of introduced Tools.

Appendix: DataWorks MCP Server

MCP (Model Context Protocol) is a standard protocol that provides context for Large Language Models (LLMs). It defines a standard way for large models to connect to different data sources and tools, thereby understanding and processing information more effectively. MCP clients can call the capabilities of various MCP Servers through the MCP protocol.

DataWorks MCP Server, as an MCP Server, encapsulates DataWorks OpenAPI and provides DataWorks big data processing capabilities. You can access DataWorks MCP Server in third-party products, programs, and Agents to quickly call DataWorks capabilities.

Important

When you are not using it in the DataWorks personal development environment, you need to configure ALIBABA_CLOUD_ACCESS_KEY_ID and ALIBABA_CLOUD_ACCESS_KEY_SECRET (please obtain them here) in the parameter env, and remove the configuration ALIBABA_CLOUD_CREDENTIALS_URI.

image