All Products
Search
Document Center

:Integrate an OpenClaw agent and configure scheduled tasks

Last Updated:Jun 09, 2026

This quickstart shows you how to integrate an OpenClaw Agent with AI Task Scheduling and create a scheduled task for automated execution.

Network architecture

Integrating an OpenClaw Agent with AI Task Scheduling involves connecting your Agent, creating a scheduled task, and verifying the setup. The network connection method depends on where your Agent is deployed:

  • Alibaba Cloud internal network — AI Task Scheduling establishes a private network connection with your VPC.

  • Public network — AI Task Scheduling accesses your Agent directly over the public network.

All network configuration and traffic costs are managed by AI Task Scheduling. No manual network setup is required.

Prerequisites

  • AI Task Scheduling is activated and a workspace is created.

  • Alibaba Cloud Container Compute Service (ACS) is activated and an ACS cluster is created.

  • A VPC, vSwitch, and security group are configured and match the ACS cluster configuration.

  • (Conditional) If you plan to import an existing Agent, the Agent is already deployed on ACS Sandbox, ECS, or another Alibaba Cloud platform.

Step 1: Integrate an Agent

Choose one of the following paths:

  • Create an Agent — You do not have an OpenClaw Agent and want AI Task Scheduling to create one using ACS Sandbox.

  • Import an Agent — You already have an Agent deployed on ACS Sandbox, ECS, or another platform and want to import it into AI Task Scheduling.

Create an Agent

Create an Agent

AI Task Scheduling uses ACS Sandbox to create the Agent runtime. Configure the following basic information and integration parameters.

Basic information

Parameter

Sub-parameter

Description

Name

The Agent name. Must be unique within the namespace.

Description

The description of the Agent.

Network Connection

Network Type

For new Agents using ACS as the runtime, only private network is supported.

VPC ID

The VPC ID. Must match the ACS cluster configuration.

VSwitch

The vSwitch. Must match the ACS cluster configuration.

Security Group

The security group. Must match the ACS cluster configuration.

Establish Connection

For the first connection to the VPC, click Establish Connection. AI Task Scheduling establishes a private network connection with your VPC.

Integration configuration

Parameter

Sub-parameter

Description

Prerequisite

Integration Method

Only ACS Sandbox is supported.

Resource Configuration

ACS Cluster

Select an ACS cluster. Create one if none exists.

Image Type

Managed Image (recommended): A customized image based on the open-source version with built-in IM channels and Agent monitoring. Custom Image: Use your own image.

Namespace

The namespace in the ACS cluster. Default value: default.

Image Address

The address of the custom image.

Image Type: Custom Image

Node Specification

Default value: 2C4G.

Model Provider

Alibaba Cloud Bailian (recommended): Use Alibaba Cloud Bailian models. Custom: Specify a custom address. Supports AI Gateway and other providers.

Model Name

The model name. Example: qwen3.6-plus.

Custom Address

The access address of the custom model provider, such as an AI Gateway address. Example: http://env-xxxxx-cn-hangzhou.alicloudapi.com/v1.

Model Provider: Custom

Model API Key

The Bailian API Key for Bailian users, or the AI Gateway API Key for AI Gateway users.

Environment Variables

Environment variable configuration for custom images.

Image Type: Custom Image

Enable Monitoring

Enable Agent monitoring. Uses CloudMonitor 2.0 for observability, which incurs charges.

Image Type: Managed Image

Service Configuration

Protocol

Select OpenAI or OpenResponse. If your OpenClaw does not have the corresponding gateway protocol enabled, enable it according to the documentation.

Gateway Token

A gateway token is automatically generated for managed images. For custom images, you must manually specify one.

Auto Scaling

Works with ACS Sandbox to automatically hibernate the Sandbox when no tasks are scheduled and wake it up when tasks are scheduled.

Import an Agent

Import an Agent

If you already have an Agent deployed on ACS Sandbox, ECS, or another Alibaba Cloud platform, you can import it into AI Task Scheduling. Configure the following basic information and integration parameters.

Basic information

Parameter

Sub-parameter

Description

Name

The Agent name. Must be unique within the namespace.

Description

The description of the Agent.

Network Connection

Network Type

Public Network: AI Task Scheduling schedules your Agent over the public network. You must allow AI Task Scheduling access through your security group. Private Network: Configure VPC settings and establish a private network connection.

VPC ID

The VPC ID. Must match the ACS cluster configuration.

VSwitch

The vSwitch. Must match the ACS cluster configuration.

Security Group

The security group. Must match the ACS cluster configuration.

Establish Connection

For the first connection to the VPC, click Establish Connection. AI Task Scheduling establishes a private network connection with your VPC.

Integration configuration

Parameter

Sub-parameter

Description

Prerequisite

Integration Method

ACS Sandbox: Integrate an existing ACS Sandbox. Manual Entry: Integrate an Agent deployed on ECS or other platforms.

Resource Configuration

ACS Cluster

Select an ACS cluster. Create one if none exists.

Integration Method: ACS Sandbox

Namespace

The namespace in the ACS cluster. Default value: default.

Integration Method: ACS Sandbox

Sandbox Name

Select a Sandbox that is already deployed on the ACS cluster.

Integration Method: ACS Sandbox

Service Configuration

Protocol

Select OpenAI or OpenResponse. If your OpenClaw does not have the corresponding gateway protocol enabled, enable it according to the documentation.

Gateway Token

Obtain the gateway token from the OpenClaw console.

Auto Scaling

Works with ACS Sandbox to automatically hibernate the Sandbox when no tasks are scheduled and wake it up when tasks are scheduled.

Integration Method: ACS Sandbox

Executor Address

Must start with http:// or https://. Domain names are supported. Example: http://192.168.0.1:18789.

Integration Method: Manual Entry

Step 2: Create a scheduled task

  1. In the Task Management menu, click Create Task.

  2. In the dialog box, select the Agent you integrated in Step 1 from the drop-down list.

  3. Configure the task parameters.

    Parameter

    Description

    Prompt

    Describe the task that the Agent will perform.

    Session

    The session mode. Task Isolation: Each run executes in an isolated session for the current task, sharing the context of the current task. Specified Session: The task runs in a specified session and shares the context of that session.

    User

    The user to run the task. If not specified, the task runs as the current task owner.

    ResponseMode

    The response mode. Supports streaming and blocking.

    Model

    The model to use. If not specified, the default model is used.

    Agent ID

    In multi-agent scenarios, specify a particular Agent. If not specified, the default main agent is used.

  4. Configure the scheduling policy. Set the time type based on your requirements:

    • none — No scheduled execution. The task can only be triggered manually.

    • cron — Recurring execution based on a cron expression.

    • one_time — One-time execution at a specified time.

  5. (Optional) Configure the notification policy by selecting notification channels, adding notification contacts, and enabling or disabling success and failure notifications.

For more information about advanced configuration, scheduling, and notifications, see the operation guide.

Step 3: Test and verify

  1. On the application management list page, click Run Once to manually trigger the task.

  2. Click Execution Records to view the task execution history.

  3. Click Logs to view the streaming output logs of the current execution.

  4. Click the details icon next to the status to view the complete output.

  5. On the Agent Management page, click the monitoring icon, or go to the Monitoring Center page to view the monitoring data of the Agent.