ACK AI Assistant is an intelligent assistant from the Alibaba Cloud Container Service team, powered by a large language model (LLM). It combines deep Kubernetes and cloud-native expertise with observability data and ACK technical expert diagnostic experience to provide intelligent diagnostics and AIOps capabilities across all phases of cluster operations.
Legal disclaimer: Before you begin, please carefully read the following notes: 1. The output of this service is generated by an LLM. Alibaba Cloud cannot guarantee the completeness or accuracy of the generated content. Use your own judgment when evaluating the output. You are solely responsible for any decisions or actions based on the generated content. 2. You are responsible for your use of the service. Ensure that the information you input is lawful and compliant, does not disrupt public order, infringe on the legal rights of others, or contain fabricated or false information. 3. This service collects status data about resources in your account (such as ECS instances and ACK clusters) to answer questions and diagnose issues. Alibaba Cloud does not collect your business data.
What ACK AI Assistant can do
ACK AI Assistant supports the following task types. For each type, sample prompts show how to phrase your requests.
Query resources
Ask questions about the state of your cluster, workloads, and resources using natural language.
Sample prompts:
-
"What is the current CPU usage of my cluster?"
-
"Show me all failed pods in the default namespace."
-
"List all nodes in my cluster."
-
"Show the YAML configuration for my nginx deployment."
-
"Get the logs for the pod named api-server-xyz."
Supported operations: Query Kubernetes resource YAML configurations, resource lists, resource details, pod logs, Kubernetes events, cluster inspection reports, and cluster lists.
Diagnose issues
Identify the root cause of cluster and application problems. ACK AI Assistant analyzes observability data—including abnormal statuses and events—and returns an explanation with a suggested fix.
Sample prompts:
-
"Why is this pod in CrashLoopBackOff?"
-
"What is causing the OOMKilled error on this node?"
-
"Diagnose the health of my cluster."
Supported operations: Diagnose cluster resources, query cluster audit logs, query Prometheus data, query Prometheus guides, and query control plane component logs.
If ACK AI Assistant cannot determine the root cause or provide a workable solution, initiate a deeper diagnosis based on ACK technical expert experience. See Use cluster diagnostics.
Manage resources
Create, update, and delete Kubernetes resources using natural language instructions.
When ACK AI Assistant attempts to create, update, or delete Kubernetes resources, it prompts for your confirmation before executing the action. Review the proposed change carefully before confirming.
Sample prompts:
-
"Create a deployment named my-app with 3 replicas using the nginx image."
-
"Scale the my-app deployment to 5 replicas."
-
"Delete the pod named stale-job-abc."
Get information
Ask general questions about Kubernetes, ACK product features, or configuration guidance.
Sample prompts:
-
"How do I configure a Horizontal Pod Autoscaler?"
-
"What are the recommended node configurations for a high-traffic workload?"
-
"Help me write a Kubernetes YAML file for a stateful application."
Supported operations: Query product knowledge, view the current time.
Query observability data
Query and analyze observability data from your cluster using natural language. ACK AI Assistant integrates LLMs with the observability best practices of Container Service for Kubernetes (ACK).
| Data source | What you can query | How it works |
|---|---|---|
| Prometheus | Metrics from Alibaba Cloud Prometheus and self-managed Prometheus instances | Converts natural language to PromQL |
| Control plane logs | Control plane Simple Log Service (SLS) logs of your ACK cluster | Runs SLS SQL queries; converts natural language to SLS SQL |
| Audit logs | Kubernetes operation audit records | Queries audit logs to audit Kubernetes operations |
For the full list of supported operations, see the alibabacloud-ack-mcp-server open-source community.
Lifecycle coverage
ACK AI Assistant covers all phases of cluster operations:
-
Day-0: Design and planning — Guidance on cloud-native containerization planning and design, including recommended configurations for clusters and nodes.
-
Day-1: Resource and application creation — Helps design and write application configuration files, such as Terraform and Kubernetes YAML files.
-
Day-2: Operations and maintenance — AIOps-based issue diagnosis, recommendations for cost and resource usage policies, and suggestions for improving high availability, stability, and security.
Prerequisites
Before you begin, ensure that you have:
-
Activated Container Service for Kubernetes and authorized the service. See Quickly create an ACK managed cluster.
-
Authorization to use the agent features of ACK AI Assistant. See Authorization for ACK AI Assistant agent features.
-
Alibaba Cloud account: No additional authorization required.
-
RAM user: Attach a permission policy that includes the
cs:CreateSessionMessageandcs:UpdateMessageFeedbackpermissions. See Manage a RAM role's permissions.
-
Open ACK AI Assistant
Open from any console page
On any page in the ACK consoleconsole, click the ACK AI Assistant floating icon to start a conversation.
Launch from intelligent diagnostics
Use this entry point when a resource anomaly is detected—for example, when a pod or workload shows an error.
-
In the ACK consoleconsole, go to the resource page with the anomaly (for example, under Workloads). Locate the abnormal resource and click Intelligent Diagnostics in the status column.
-
ACK AI Assistant analyzes the anomaly and provides a diagnosis explaining the likely cause and recommended action. If the diagnosis is inconclusive, select the option to run a deeper analysis. ACK AI Assistant guides you through additional steps—such as confirming the target resource—and returns a detailed report with suggested solutions.
Manage conversation history
Each Q&A or diagnostics session opens a new conversation. Recent messages are saved as context for subsequent responses.
To start a fresh conversation, refresh the page or click Clear history.
What's next
-
Use cluster diagnostics — Run a deeper, expert-guided diagnosis for abnormal pods or nodes.
-
Observability best practices for ACK — Learn about the observability data sources ACK AI Assistant queries.
Contact us
If you have feedback or suggestions, join our DingTalk group (ID: 70080006301).