CMS_GetCpuloadavg15mData
Return the fifteen-minute CPU load average
Start, stop and re-key ECS and RDS instances in batches, read their CloudMonitor metrics, and push a build onto them — all addressed by instance ID list and region.
LABSCO SUMMARY
The OOS tools take a list of instance IDs, so starting a group of machines, rotating their passwords or swapping their system disks is a single call. Metrics, inventory and deployment sit behind the same access key, which puts the question — is this instance loaded — and the action — restart it — in the same session.
Alibaba Cloud operations in 40 tools: OOS batch actions over ECS and RDS, CloudMonitor metric reads, OSS bucket management, ECS/VPC/RDS inventory queries, and three tools that act on the local machine to prepare a deployment.
An Alibaba Cloud access key — ALIBABA_CLOUD_ACCESS_KEY_ID and ALIBABA_CLOUD_ACCESS_KEY_SECRET — and uv on your PATH, which resolves the Python runtime through UV_PYTHON. LOCAL_RunShellScript runs on the machine hosting the server, not on an ECS instance.
One command plus a key — uvx alibaba-cloud-ops-mcp-server@latest, then supply credentials
USE CASES
Find the instances in the region, reboot them as one batch rather than one at a time, and watch CPU utilization come back to check the change took.
Read the load averages at one, five and fifteen minutes alongside memory and disk utilization for the same instances, so a spike can be separated from a machine that has been saturated all week.
Identify what kind of project it is, deploy the built artifact to the target instances through OOS CodeDeploy, then read the application group's deployment status rather than guessing from the logs.
Confirm the instance type and system disk you want are actually available in the target zone, and that the account quota allows another one, before the create call fails halfway.
Select the instances, change the password on all of them in one operation, and plan for the reboot the operation carries with it.
List the compute, the networks it sits on, the databases behind it and the buckets it reads, so the move is planned against what is there rather than what the diagram says.
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
CMS_GetCpuloadavg15mDataReturn the fifteen-minute CPU load average
CMS_GetCpuloadavg5mDataReturn the five-minute CPU load average
CMS_GetCpuLoadavgDataReturn the one-minute CPU load average
CMS_GetCpuUsageDataReturn CPU utilization data for ECS instances
CMS_GetDiskTotalDataReturn total disk partition capacity metric data
CMS_GetDiskUsageDataReturn disk utilization metric data
CMS_GetDiskUsedDataReturn disk partition used-space metric data
CMS_GetMemUsageDataReturn memory utilization metric data
CMS_GetMemUsedDataReturn memory-used metric data
ECS_DeleteInstancesRelease pay-as-you-go instances or expired subscription instances in a batch, with a parameter deciding whether disks are released or kept as pay-as-you-go
ECS_DescribeAccountAttributesReturn your ECS quotas in a region: security groups, elastic network interfaces, pay-as-you-go and spot vCPU counts, total pay-as-you-go disk capacity, dedicated hosts, network type, and whether the account has completed real-name verification
ECS_DescribeAvailableResourceCheck resource stock in a zone before creating an instance or changing an instance specification
ECS_DescribeImagesQuery the images available to you, filtered by image ID, usage scenario or other filters
ECS_DescribeInstancesQuery the instance list by a range of conditions, with each instance's detail attached
ECS_DescribeRegionsQuery the list of regions, filtered by billing method and resource type
ECS_DescribeSecurityGroupsQuery security groups by region, security group ID or security group type
ECS_DescribeZonesQuery the availability zones of a region, filtered by billing method
LOCAL_AnalyzeDeployStackIdentify a local project's deployment method and stack, covering npm, Python, Java, Go and Docker
LOCAL_ListDirectoryList the files and subdirectories of a local directory, with name, type and size
LOCAL_RunShellScriptRun a shell script or command locally and return its output, error output and exit code
OOS_CodeDeployDeploy an application to ECS instances through the OOS CodeDeploy feature, from a build artifact placed in .code_deploy/release
OOS_GetDeployStatusReturn the deployment status of an application group
OOS_GetLastDeploymentInfoReturn the details of the previous application deployment
OOS_RebootInstancesReboot a batch of ECS instances in one call
OOS_RebootRDSInstancesReboot a batch of RDS instances in one call
OOS_ReplaceSystemDiskReplace the system disk on a batch of ECS instances, changing their operating system
OOS_ResetPasswordChange the password on a batch of ECS instances; the instances are rebooted as part of the operation
OOS_RunCommandRun a Cloud Assistant command across many ECS instances at once, for work such as application management and resource tagging
OOS_RunInstancesCreate a batch of ECS instances in one call, with image, instance type, disks, network and tags given per request
OOS_StartInstancesStart a batch of ECS instances in one call
OOS_StartRDSInstancesStart a batch of RDS instances in one call
OOS_StopInstancesStop a batch of ECS instances in one call
OOS_StopRDSInstancesStop a batch of RDS instances in one call
OSS_DeleteBucketDelete an OSS bucket
OSS_ListBucketsList the OSS buckets in a region
OSS_ListObjectsList the objects held in an OSS bucket
OSS_PutBucketCreate an OSS bucket
RDS_DescribeDBInstancesQuery the list of RDS instances
VPC_DescribeVpcsQuery the VPCs that have been created
VPC_DescribeVSwitchesQuery vSwitches, the unit the internal network is organised by
BEFORE YOU INSTALL
40 tools when we connected. Registered names carry a product prefix (OOS_RunCommand, not RunCommand), and StartDBInstances, StopDBInstances and RestartDBInstances are OOS_StartRDSInstances, OOS_StopRDSInstances and OOS_RebootRDSInstances. ECS_DescribeAccountAttributes is missing from the documentation table.
CONTEXT COST
Rated by the size of the 40 tool definitions that ride along on every turn, not by how many there are
INSTALLATION
uvx alibaba-cloud-ops-mcp-server@latest{
"mcpServers": {
"alibaba-cloud-ops-mcp-server": {
"timeout": 600,
"command": "uvx",
"args": [
"alibaba-cloud-ops-mcp-server@latest"
],
"env": {
"ALIBABA_CLOUD_ACCESS_KEY_ID": "<YOUR_ALIBABA_CLOUD_ACCESS_KEY_ID>",
"ALIBABA_CLOUD_ACCESS_KEY_SECRET": "<YOUR_ALIBABA_CLOUD_ACCESS_KEY_SECRET>"
}
}
}
}Install link: qoder://aicoding.aicoding-deeplink/mcp/add?name=alibaba-cloud-ops-mcp-server&config=JTdCJTIyY29tbWFuZCUyMiUzQSUyMnV2eCUyMiUyQyUyMmFyZ3MlMjIlM0ElNUIlMjJhbGliYWJhLWNsb3VkLW9wcy1tY3Atc2VydmVyJTQwbGF0ZXN0JTIyJTVEJTJDJTIyZW52JTIyJTNBJTdCJTIyQUxJQkFCQV9DTE9VRF9BQ0NFU1NfS0VZX0lEJTIyJTNBJTIyWW91ciUyMEFjY2VzcyUyMEtleSUyMElkJTIyJTJDJTIyQUxJQkFCQV9DTE9VRF9BQ0NFU1NfS0VZX1NFQ1JFVCUyMiUzQSUyMllvdXIlMjBBY2Nlc3MlMjBLZXklMjBTRUNSRVQlMjIlN0QlN0Q%3D
Install link: https://cursor.com/en/install-mcp?name=alibaba-cloud-ops-mcp-server&config=eyJ0aW1lb3V0Ijo2MDAsImNvbW1hbmQiOiJ1dnggYWxpYmFiYS1jbG91ZC1vcHMtbWNwLXNlcnZlckBsYXRlc3QiLCJlbnYiOnsiQUxJQkFCQV9DTE9VRF9BQ0NFU1NfS0VZX0lEIjoiWW91ciBBY2Nlc3MgS2V5IElkIiwiQUxJQkFCQV9DTE9VRF9BQ0NFU1NfS0VZX1NFQ1JFVCI6IllvdXIgQWNjZXNzIEtleSBTZWNyZXQifX0%3D