ADBAgent

Business operations analysis

Operations teams face long turnaround times when requesting data from analysts. ADBAgent lets analysts configure analysis scenarios linked to core tables and Skills, enabling business users to self-serve routine queries in natural language. Results can be scheduled as daily reports and delivered to DingTalk or Feishu.

Cross-department data collaboration

ADBAgent centralizes data scenarios and Skills so all teams use consistent metrics. Shared scenarios prevent redundant work, and full traceability simplifies metric discrepancy troubleshooting.

Automated business decision-making

Integrate ADBAgent with OpenClaw through a Skill to trigger recurring analysis tasks and run optimization strategies automatically. For example, trigger restocking on low-stock alerts or launch retention campaigns based on churn predictions—closing the loop from data insight to action.

Step 1: Create a cluster

1. Log on to the AnalyticDB for MySQL console. In the upper-right corner of the page, click Buy Cluster.

2. Configure the following core parameters on the purchase page. Use defaults for other parameters. All parameters are described in Create a cluster.

   • Product Type: Select Pay-as-you-go.

   • Edition: Select Enterprise Edition (Multi-replica).

   • Deployment Mode: Select Single-AZ deployment.

   • Region and Primary Zone: ADBAgent is currently available only in the China (Hangzhou), China (Shanghai), and China (Beijing) regions.

   • Auto-enable PVL service: Select this check box.

   • Number of Reserved Resource Nodes: Set this parameter based on your requirements. If you set this parameter to 0, you can create and access only external tables. You cannot create or access internal tables.

3. Click Buy Now and follow the instructions on the console to complete the purchase.

   Cluster creation takes 10–15 minutes after payment. The cluster is ready when its status changes to Running on the Clusters page.

Step 2: Create a database account

ADBAgent connects to an AnalyticDB for MySQL cluster through a database account. Account types differ in permissions.

1. On the Clusters page, click the ID of your target cluster to go to the cluster details page.

2. In the left-side navigation pane, click Accounts.

3. On the Database Accounts tab, click Create Account. Create the following two types of database accounts:

   1. Privileged account: Used to manage the permissions of standard accounts.

   2. Standard account: This account is for the ADBAgent database connection.

4. Configure permissions for the standard account.

   In the row of the standard account, click the ⋮ icon in the Actions column and select Edit Permissions. We recommend that you grant the SELECT permission.

Step 3: Initialize ADBAgent

1. In the left-side navigation pane of the cluster details page, click ADBAgent (AI Assistant).

2. Click Get Started.

   On first access, ADBAgent initializes automatically. This takes about 2 minutes.

Step 4: Configure a whitelist

To access ADBAgent, make the service publicly accessible and configure a whitelist for authorized IP addresses.

1. On the ADBAgent (AI Assistant) page, click Add whitelist group in the upper-right corner of the Whitelist settings section.

   

2. In the Add whitelist group dialog box, set the following parameters.

   • Group Name: Enter a custom name.

   • IP addresses in whitelist: Enter the public IP addresses or CIDR blocks of the machines that need to access ADBAgent. To find the egress IP address of your machine, do one of the following:

     • Run the following shell command: curl ifconfig.co/json

     • In a web browser, visit https://ipconfig.com/zh

3. Click OK.

Step 5 (Optional): Enable public network access

To use the following ADBAgent features, create a NAT Gateway to enable public network access:

• Channel integration: Connect to messaging platforms like DingTalk or Feishu.

• Intelligent diagnostics: Use the AnalyticDB for MySQL AI diagnostic Skill to quickly diagnose your cluster.

• Public data source: Connects to a data source over the public network.

Procedure:

1. Obtain the VPC and vSwitch information of the AnalyticDB for MySQL cluster.

   1. Log on to the AnalyticDB for MySQL console. In the upper-left corner of the console, select a region. In the left-side navigation pane, click Clusters. Find the cluster that you want to manage and click the cluster ID.

   2. In the left-side navigation pane, click Cluster Management > Cluster Information.

   3. In the Network Information section, find the VPC ID and vSwitch ID.

2. Create a NAT Gateway and bind an EIP to it.

   1. Go to the NAT Gateway purchase page.

   2. Configure the NAT Gateway.

      • Region: Select the region where your AnalyticDB for MySQL cluster is located.

      • Network and Zone: Select the VPC and vSwitch that you identified in the previous step.

      • EIP: If you already have an EIP, select Select EIP. Otherwise, select Purchase EIP.

   3. Click Buy Now.

3. Configure an SNAT entry.

   1. Go to the NAT Gateway console. In the left-side navigation pane, select NAT Gateway > Internet NAT Gateway, and then click the newly created NAT gateway.

   2. Select the SNAT tab and click Create SNAT Entry.

   3. Select the EIP and click OK.

Step 1: Access ADBAgent

1. On the ADBAgent (AI Assistant) page, find the target service, click Credentials in the Actions column, and then view the Default Account and Default Password.

2. Click Access and log on to ADBAgent with the default account and password to get started.

Step 2: Configure a data source

On the ADBAgent page, click Data Connection in the left-side navigation pane to configure a data source.

ADBAgent supports multiple data source types.

Step 3: Start an analysis task

In the left-side navigation pane, click New Task. Describe your analysis requirements and send the request. ADBAgent interprets the intent, matches Skills, explores data in layers, runs SQL, and presents structured results.

• Specify a scenario and Skills (Optional)

  Click + in the input box to select a scenario, or enter /Skill name to call a Skill directly. The task runs within the corresponding data scope and analysis logic.

  

• Multi-turn conversation

  Complex data analysis scenarios support multi-turn conversations:

  • Follow-up analysis: Ask follow-up questions based on the results to dig deeper into the data.

  • Error correction: If the analysis results are not as expected, directly point out the problem. The system automatically adjusts the analysis logic.

  • New requirements: Add new analysis dimensions or filter conditions at any time during the analysis process.

Step 4: Save a task as a Skill

Save an analysis process as a reusable Skill.

In an analysis task, prompt ADBAgent with something like "Summarize this task as a Skill and add it to my Skills" to save the analysis logic. Call the saved Skill directly in future tasks.

Step 5: Save as an analysis scenario

Save frequently used analysis workflows as reusable scenarios for consistent reproduction.

1. In the left-side navigation pane, click Business scenarios.

2. Click Create scenario Plugin, and then enter a name and description for the scenario.

3. Associate the target data table from the database.

4. Bind the required Skills.

5. After saving, select this scenario when creating tasks to apply its analysis scope and Skill set.

Skill management

In the left-side navigation pane, click Skill library to manage analysis Skills.

• Interactive creation: Click Create your Skill and describe your requirements in a conversation. The system generates Skills automatically.

  

• Content fine-tuning: Edit Skill content to adjust the analysis logic.

• Public skill market: Browse and filter community Skills by category.

  

OpenClaw integration

Integrate ADBAgent with OpenClaw to add data analysis to your automation workflows.

Three-layer data exploration

Industry-specific data analysis

ADBAgent includes built-in Skills for ad marketing and performance attribution, such as conversion rate analysis and campaign diagnostics—a diagnostic system from macro trends to micro root causes, and then to a closed-loop strategy: