All Products
Search
Document Center

Security Center:Enable and disable Agentic Agentless Detection

Last Updated:Oct 29, 2026

Agentic Agentless Detection performs security checks on servers without installing an agent, and you can enable or disable the feature as needed.

What is Agentic Agentless Detection

The Agentless Detection feature allows you to assess the security risks of cloud servers without installing the Security Center agent. The feature scans the images of your servers, uses AI sandbox-based dynamic analysis, and performs multi-dimensional security checks in an isolated environment, including checks for vulnerabilities, malware, configuration baselines, and sensitive files. The feature also supports AI-powered vulnerability fixing.

How it works

  1. Create a full-server image (host scan tasks only): Based on the task configurations, the system creates a full-server image for the disks of the target Elastic Compute Service (ECS) instance.

  2. Share and mount: The system shares the created snapshot or image with the dedicated analysis cluster of Security Center.

  3. Isolated scanning: In an isolated dedicated environment, the analysis engine mounts the file system of the snapshot or image and performs a security scan. The AI sandbox detonates suspicious files or images and analyzes their runtime behaviors to improve the accuracy of threat analysis. This process consumes no computing resources of the target server.

  4. Report generation and cleanup: After the scan is complete, the system generates a risk report and automatically deletes the temporary image based on the configured policies to save storage costs.

Common use cases

  • "Zero-interference" risk assessment: Perform non-intrusive security risk assessment that consumes no resources on core production systems that cannot run an agent or cannot tolerate performance impact, to ensure business continuity.

  • Unified cross-platform asset security detection: Cover all asset types, including legacy and proprietary systems, and quickly assess the overall security posture of a cross-platform environment from a unified view.

  • Visualized multi-dimensional risk detection: A single scan detects multi-dimensional risks such as vulnerabilities, malicious files, configuration baselines, and sensitive information, and visualizes the overall security posture.

  • Asset compliance and security review: Before you create instances or launch services, review the security of custom images and host snapshots to ensure that the delivered production environment meets security and compliance standards.

  • AI-powered automatic fixing loop: Based on the AI sandbox analysis results, the feature provides one-click or automated fixing suggestions and execution for risks that can be fixed, such as specific application vulnerabilities, baseline alerts, and malicious file alerts, to create a closed loop from risk detection to remediation.

Enable Agentic Agentless Detection

  1. Enable the service

    1. Log on to Security Center console.Select the region where the asset to be protected is located: Chinese Mainland or Outside Chinese Mainland.

    2. On the Overview page, turn on the Agentic Agentless Detection switch in the Enable Pay-as-You-Go Service section.

    3. On the enablement page, confirm the billing information (Agentless detection and analysis are billed by scanned data volume; vulnerability fixing is billed by the number of successfully fixed vulnerabilities. Both are billed daily and can be unsubscribed at any time). By default, Enable Policy (Recommended) is selected. Click Enable Now and Authorize.

      Important

      Agentless detection uses the pay-as-you-go billing method.

    4. You can also go to the Protection Configuration > Host Protection > Agentless Detection page and click Activate Now and select the relevant agreement.

  2. Complete service authorization (first time)

    When you use the feature for the first time, the system prompts you to authorize a service role. Follow the on-screen instructions and click Authorize Now.

    Note

    After the authorization is successful, Security Center automatically creates the service-linked roles AliyunServiceRoleForSas and AliyunServiceRoleForSasAgentless. For more information about these service-linked roles, see Service-linked roles for Security Center.

  3. Authorize Agentless Detection for encrypted ECS disks (first time)

    When you use the feature for the first time, if your ECS instances use encrypted disks (encrypted system disks or encrypted data disks) and you need to perform agentless detection on them, complete the authorization by following the instructions in Authorize for ECS encrypted disks.

Disable Agentic Agentless Detection

  • Disable entry:

    • On the Overview page of the Security Center console, turn off the Agentic Agentless Detection switch in the Enable Pay-as-You-Go Service section.

    • On the Agentless Detection page of the Security Center console, click Suspended in the Risk Detection section.

  • Data cleanup:

    • The results of detection tasks are retained for only 30 days and are automatically cleared after expiration.

    • Scan task configurations are retained and not cleared.

Billing

The following describes the billing for the Agentless Detection feature:

  • Scan and fixing fees (Agentless Detection)

    • Billing method: Pay-as-you-go.

    • Billing cycle: Natural day.

    • Unit price:

      • Agentic agentless detection and analysis: USD 0.13/GB.

      • Agentic vulnerability fixing: USD 1.33/vulnerability.

    • Billing rules:

      • Agentic Agentless Detection and Analysis: Calculated based on the actual data volume of the scanned images, not the total disk capacity.

      • Agentic Vulnerability Fixing: Charged by the number of successfully fixed vulnerabilities, not by the number of tasks or attempts.

        • If a fixing task involves three vulnerabilities and all are fixed, it counts as three successfully fixed vulnerabilities.

      • Agentic Baseline Check Fixing and Malicious File Fixing: Charged by the number of successfully fixed alerts, not by the number of tasks or attempts.

  • ECS resource usage fees

    Important

    We recommend that you select Retain Only At-risk Image when you configure a host scan task. The system automatically deletes risk-free images after the scan is complete to save storage costs. For more information, see Retention Period Configuration.

    • Image fees: Detection tasks create images for servers, which are charged based on the image usage capacity and duration. These fees are charged by ECS. For more information, see Image billing.

    • Encrypted disk scan resource fees: To scan ECS encrypted disks, the system creates the following resources under your account each time a scan is performed. Some resources incur minimal fees and are released immediately after the scan.

      • ECS instance: When you select Service Key encryption for the ECS encrypted disk to be scanned, we create a preemptive ECS instance for encrypted disk scanning. These fees are charged by ECS. For more information, see Instance type billing.

      • VPC instance: A VPC instance named alibaba-cloud-security-scan-vpc. No fees are incurred for creating this instance.

      • VSwitch instance: A VSwitch instance named alibaba-cloud-security-scan-subnet. No fees are incurred for creating this instance.

FAQ

  • What is the difference between Agentless Detection and Virus Blocking?

    Comparison item

    Agentless Detection

    Virus Blocking

    Working mode

    Scans offline snapshots and uses AI sandbox-based dynamic analysis

    Real-time online monitoring and scanning, dynamic and static analysis

    Server status

    Can detect servers that are started or stopped

    Can detect only servers that are started and have the agent online

    Detection scope

    Vulnerabilities, baselines, malicious samples, and sensitive files

    Viruses, webshells, intrusions, vulnerabilities, and more

    Handling capability

    Supports detection, alerting, and whitelisting, and provides AI automatic fixing or one-click fixing for some risks

    Provides one-click isolation, blocking, and fixing

    Performance impact

    None

    Minimal (the agent consumes a small amount of system resources)

    Billing mode

    Pay-as-you-go (by scanned GB)

    Subscription (Anti-virus edition or higher) or pay-as-you-go

    Scan mode

    Supports full-disk scan

    Supports quick scan and custom directory scan

  • Can Agentless Detection automatically fix all risks?

    No, it cannot automatically fix all risks. It supports AI automatic fixing for only some application vulnerabilities. For the list of supported vulnerabilities, see Vulnerability fix support list.

    Note
    • Fixing is completed by creating a new custom security image. The original image is not modified, ensuring business continuity and rollback capability.

    • For risks that do not support automatic fixing, follow the handling suggestions in the risk details page to manually complete the fixing.

  • How do I use the advanced detection feature for non-Alibaba Cloud servers?

    Currently, Agentless Detection supports access to Alibaba Cloud ECS, AWS EC2, and Azure servers.

    Important
    • Azure is supported for access and detection only in Outside Chinese Mainland. For more information, see Detect and fix risks.

    • For servers from other cloud providers or on-premises data centers, we recommend that you install the Security Center agent and enable the Anti-virus edition or a higher edition to obtain comprehensive security protection capabilities. For more information, see Purchase Security Center and Install the agent.

  • Why is the asset IP not displayed for some Agentless Detection alerts?

    When you use Agentless Detection, if the asset IP is empty, the possible causes are as follows:

    • The instance is released

      After the asset instance corresponding to the alert is released, Security Center cannot obtain the IP address of the instance because the instance no longer exists in the system and its IP address cannot be tracked or displayed.

    • Data cleanup mechanism

      Security Center may clean up the records of released instances, causing the related information to fail to load.

  • Does Agentless Detection support on-premises IDCs?

    No. Agentless Detection relies on the cloud image mechanism for offline analysis. The system automatically creates images for specified servers to perform risk detection. On-premises servers cannot create cloud images. To perform security detection on on-premises servers, install the Security Center agent and use agent-based detection features such as Virus Blocking and Vulnerability Scan.

Supported regions

Agentic agentless detection is supported in the following regions and features:

Alibaba Cloud

regionId

Region name

Standard/encrypted disk scan

Parallel sandbox

Vulnerability fixing

cn-hangzhou

China (Hangzhou)

Supported

Supported

Unsupported

cn-qingdao

China (Qingdao)

Supported

Supported

Unsupported

ap-northeast-1

Japan (Tokyo)

Supported

Supported

Unsupported

ap-northeast-2

South Korea (Seoul)

Supported

Supported

Unsupported

ap-southeast-1

Singapore

Supported

Supported

Unsupported

ap-southeast-3

Malaysia (Kuala Lumpur)

Supported

Supported

Unsupported

ap-southeast-5

Indonesia (Jakarta)

Supported

Supported

Unsupported

ap-southeast-6

Philippines (Manila)

Supported

Supported

Unsupported

ap-southeast-7

Thailand (Bangkok)

Supported

Supported

Unsupported

ap-southeast-8

Malaysia (Johor)

Supported

Supported

Unsupported

cn-beijing

China (Beijing)

Supported

Supported

Unsupported

cn-chengdu

China (Chengdu)

Supported

Supported

Unsupported

cn-guangzhou

China (Guangzhou)

Supported

Supported

Unsupported

cn-heyuan

China (Heyuan)

Supported

Supported

Unsupported

cn-hongkong

China (Hong Kong)

Supported

Supported

Unsupported

cn-huhehaote

China (Hohhot)

Supported

Supported

Unsupported

cn-shanghai

China (Shanghai)

Supported

Supported

Unsupported

cn-shenzhen

China (Shenzhen)

Supported

Supported

Unsupported

cn-wuhan-lr

China (Wuhan) (Local Region)

Supported

Supported

Unsupported

cn-wulanchabu

China (Ulanqab)

Supported

Supported

Unsupported

cn-zhangjiakou

China (Zhangjiakou)

Supported

Supported

Unsupported

cn-zhongwei

China (Zhongwei)

Supported

Supported

Unsupported

eu-central-1

Germany (Frankfurt)

Supported

Supported

Unsupported

eu-west-1

UK (London)

Supported

Supported

Unsupported

eu-west-2

France (Paris)

Supported

Unsupported

Unsupported

eu-west-3

Netherlands (Amsterdam)

Supported

Unsupported

Unsupported

me-central-1

Saudi Arabia (Riyadh)

Supported

Unsupported

Unsupported

me-east-1

UAE (Dubai)

Supported

Supported

Unsupported

na-south-1

Mexico

Supported

Supported

Unsupported

sa-east-1

Brazil (Sao Paulo)

Supported

Supported

Unsupported

us-east-1

US (Virginia)

Supported

Supported

Unsupported

us-west-1

US (Silicon Valley)

Supported

Supported

Unsupported

AWS

China site

regionId

Region name

Standard/encrypted disk scan

Parallel sandbox

Vulnerability fixing

cn-north-1

China (Beijing)

Supported

Unsupported

Unsupported

cn-northwest-1

China (Ningxia)

Supported

Unsupported

Unsupported

International site

regionId

Region name

Standard/encrypted disk scan

Parallel sandbox

Vulnerability fixing

af-south-1

Africa (Cape Town)

Supported

Unsupported

Unsupported

ap-east-1

Asia Pacific (Hong Kong)

Supported

Unsupported

Unsupported

ap-east-2

Asia Pacific (Taipei)

Supported

Unsupported

Unsupported

ap-northeast-1

Asia Pacific (Tokyo)

Supported

Unsupported

Unsupported

ap-northeast-2

Asia Pacific (Seoul)

Supported

Unsupported

Unsupported

ap-northeast-3

Asia Pacific (Osaka)

Supported

Unsupported

Unsupported

ap-south-1

Asia Pacific (Mumbai)

Supported

Unsupported

Unsupported

ap-south-2

Asia Pacific (Hyderabad)

Supported

Unsupported

Unsupported

ap-southeast-1

Asia Pacific (Singapore)

Supported

Unsupported

Unsupported

ap-southeast-2

Asia Pacific (Sydney)

Supported

Unsupported

Unsupported

ap-southeast-3

Asia Pacific (Jakarta)

Supported

Unsupported

Unsupported

ap-southeast-4

Asia Pacific (Melbourne)

Supported

Unsupported

Unsupported

ap-southeast-5

Asia Pacific (Malaysia)

Supported

Unsupported

Unsupported

ap-southeast-6

Asia Pacific (New Zealand)

Supported

Unsupported

Unsupported

ap-southeast-7

Asia Pacific (Thailand)

Supported

Unsupported

Unsupported

ca-central-1

Canada (Central)

Supported

Unsupported

Unsupported

ca-west-1

Canada West (Calgary)

Supported

Unsupported

Unsupported

eu-central-1

Europe (Frankfurt)

Supported

Unsupported

Unsupported

eu-central-2

Europe (Zurich)

Supported

Unsupported

Unsupported

eu-north-1

Europe (Stockholm)

Supported

Unsupported

Unsupported

eu-south-1

Europe (Milan)

Supported

Unsupported

Unsupported

eu-south-2

Europe (Spain)

Supported

Unsupported

Unsupported

eu-west-1

Europe (Ireland)

Supported

Unsupported

Unsupported

eu-west-2

Europe (London)

Supported

Unsupported

Unsupported

eu-west-3

Europe (Paris)

Supported

Unsupported

Unsupported

il-central-1

Israel (Tel Aviv)

Supported

Unsupported

Unsupported

me-central-1

Middle East (UAE)

Supported

Unsupported

Unsupported

me-south-1

Middle East (Bahrain)

Supported

Unsupported

Unsupported

mx-central-1

Mexico (Central)

Supported

Unsupported

Unsupported

sa-east-1

South America (Sao Paulo)

Supported

Unsupported

Unsupported

us-east-1

US East (N. Virginia)

Supported

Unsupported

Unsupported

us-east-2

US East (Ohio)

Supported

Unsupported

Unsupported

us-west-1

US West (N. California)

Supported

Unsupported

Unsupported

us-west-2

US West (Oregon)

Supported

Unsupported

Unsupported

Azure

China site

regionId

Region name

Standard/encrypted disk scan

Parallel sandbox

Vulnerability fixing

chinanorth3

China North 3

Supported

Unsupported

Unsupported

International site

regionId

Region name

Standard/encrypted disk scan

Parallel sandbox

Vulnerability fixing

australiacentral

Australia Central

Supported

Unsupported

Unsupported

australiaeast

Australia East

Supported

Unsupported

Unsupported

australiasoutheast

Australia Southeast

Supported

Unsupported

Unsupported

austriaeast

Austria East

Supported

Unsupported

Unsupported

belgiumcentral

Belgium Central

Supported

Unsupported

Unsupported

brazilsouth

Brazil South

Supported

Unsupported

Unsupported

canadacentral

Canada Central

Supported

Unsupported

Unsupported

canadaeast

Canada East

Supported

Unsupported

Unsupported

centralindia

Central India

Supported

Unsupported

Unsupported

centralus

Central US

Supported

Unsupported

Unsupported

chilecentral

Chile Central

Supported

Unsupported

Unsupported

eastasia

East Asia (Hong Kong)

Supported

Unsupported

Unsupported

eastus

East US

Supported

Unsupported

Unsupported

eastus2

East US 2

Supported

Unsupported

Unsupported

francecentral

France Central

Supported

Unsupported

Unsupported

germanywestcentral

Germany West Central

Supported

Unsupported

Unsupported

indonesiacentral

Indonesia Central

Supported

Unsupported

Unsupported

israelcentral

Israel Central

Supported

Unsupported

Unsupported

italynorth

Italy North

Supported

Unsupported

Unsupported

japaneast

Japan East

Supported

Unsupported

Unsupported

japanwest

Japan West

Supported

Unsupported

Unsupported

koreacentral

Korea Central

Supported

Unsupported

Unsupported

koreasouth

Korea South

Supported

Unsupported

Unsupported

malaysiawest

Malaysia West

Supported

Unsupported

Unsupported

mexicocentral

Mexico Central

Supported

Unsupported

Unsupported

newzealandnorth

New Zealand North

Supported

Unsupported

Unsupported

northcentralus

North Central US

Supported

Unsupported

Unsupported

northeurope

North Europe (Ireland)

Supported

Unsupported

Unsupported

norwayeast

Norway East

Supported

Unsupported

Unsupported

polandcentral

Poland Central

Supported

Unsupported

Unsupported

qatarcentral

Qatar Central

Supported

Unsupported

Unsupported

southafricanorth

South Africa North

Supported

Unsupported

Unsupported

southcentralus

South Central US

Supported

Unsupported

Unsupported

southeastasia

Southeast Asia (Singapore)

Supported

Unsupported

Unsupported

southindia

South India

Supported

Unsupported

Unsupported

spaincentral

Spain Central

Supported

Unsupported

Unsupported

swedencentral

Sweden Central

Supported

Unsupported

Unsupported

switzerlandnorth

Switzerland North

Supported

Unsupported

Unsupported

uaenorth

UAE North

Supported

Unsupported

Unsupported

uksouth

UK South

Supported

Unsupported

Unsupported

ukwest

UK West

Supported

Unsupported

Unsupported

westcentralus

West Central US

Supported

Unsupported

Unsupported

westeurope

West Europe (Netherlands)

Supported

Unsupported

Unsupported

westindia

West India

Supported

Unsupported

Unsupported

westus

West US

Supported

Unsupported

Unsupported

westus2

West US 2

Supported

Unsupported

Unsupported

westus3

West US 3

Supported

Unsupported

Unsupported

Tencent Cloud

regionId

Region name

Standard/encrypted disk scan

Parallel sandbox

Vulnerability fixing

ap-guangzhou

China (Guangzhou)

Supported

Unsupported

Unsupported