OceanBase OAgent provides built-in skills and other skills to extend intelligent O&M capabilities. You can view and manage skills in the OceanBase OAgent settings as needed.
Skills are automatically triggered in specific scenarios without manual invocation. When your query or operation matches the trigger conditions of a skill, OceanBase OAgent automatically invokes the corresponding skill to provide services for you.
View skills
- In the left-side navigation pane, select OceanBase OAgent.
- On the OceanBase OAgent page, click Settings.
- On the Settings page, select the Skills tab. You can view the list of currently supported skills, such as healthcheck, mcporter, and node-connect.
Introduction to OceanBase skills
The following table provides an overview of currently supported skills. Each skill is automatically invoked when its trigger scenario is matched, without manual specification.
alarm-report
Description: Generates a global alert report for OceanBase Cloud. Based on the GetAlarmHistoryList interface of the obcloud-api skill, it aggregates alert history by organization/project and presents the results in a table format (high-risk alerts first).
Trigger scenarios: View alerts, alert status, check for alerts, generate an alert report, alert reporting, recent P0 alerts, project-level alert summary, etc.
Usage mode: Supports two modes: global scan (automatically enumerates all organizations/projects) and specified project.
capacity-forecast
Description: Performs capacity forecasting for OceanBase Cloud instances. Based on the DescribeInstanceCapacityProfile and DescribeMetricsData interfaces, it analyzes disk/CPU capacity trends for an instance, outputs the number of available disk days, resource growth trends, and expansion recommendations.
Trigger scenarios: Capacity forecasting, capacity analysis, how much storage is left, disk is almost full, how much longer can it last, whether expansion is needed, expansion assessment, capacity planning, resource growth trend, storage prediction, CPU trend, pull a trend, long-term load trend, CPU usage in the past few days, disk growth, capacity report, etc. (English expressions are also supported, such as capacity forecast, disk almost full, capacity planning, etc.).
Usage mode: Supports an overview of all instances (Lightweight mode) and in-depth analysis of a specified instance (Deep mode).
greptimedb-sql-audit
Description: In-depth analysis of OceanBase SQL audit logs based on GreptimeDB. By using mcporter to call the execute_sql tool of greptimedb MCP, it performs in-depth diagnostics on known sql_id for execution plan changes, abnormal resource consumption, and traffic imbalance.
Application note: This skill is used in the "In-depth analysis" phase of SQL diagnostics, not the "Detection" phase. It is suitable for scenarios requiring in-depth analysis of the execution details of a specific SQL statement (such as plan jitter, large/small accounts, hard parsing, lock contention, etc.). To simply view the TopSQL/SlowSQL list, use the obcloud-api skill.
Trigger scenarios: Analyze this SQL, SQL execution plan changes, sql_audit in-depth query, why is this SQL slow, SQL traffic distribution, hard parsing analysis, large/small accounts, etc.
instance-load-query
Description: Queries the load of an OceanBase Cloud instance. Based on the DescribeInstanceCapacityProfile interface of the obcloud-api skill, it aggregates resource metrics such as the current CPU load, disk usage rate, and connection count ratio of the instance.
Trigger scenarios: Check the load, instance load, cluster status, cluster load, resource usage, which tenant has the highest resource usage, what is the CPU usage, how much disk space is left, is the connection count high, help me check the status of all clusters, instance resource metrics, etc. (English expressions are also supported, such as instance load, cluster load, resource usage, etc.).
Usage mode: Supports an overview of all instances (Lightweight mode) and in-depth analysis of a specified instance (Deep mode).
ob-health-check
Description: Provides a comprehensive health check for OceanBase Database, covering resource assessment, performance monitoring, SQL diagnostics, alert history, and backup status. It can automatically generate inspection reports (in Markdown format by default, with support for conversion to PDF) and supports scheduled execution and IM notifications (Feishu/DingTalk).
Trigger scenarios: Database inspection, health check, database checkup, OB inspection, cluster diagnosis, SQL inspection, SQL diagnostics, etc.
obcloud-api
Description: The OceanBase Cloud OpenAPI documentation. Use this when you need to query or manage resources such as clusters, tenants, databases, users, alerts, backups, SQL diagnostics, connection addresses, security configurations, monitoring metrics, tags, and parameters in the cloud.
Trigger scenarios: Even if the user does not explicitly mention "call the API," the scenario should trigger if the described operation can be mapped to an OceanBase Cloud API. For example: "Check the status of this cluster," "What alerts have occurred recently?" "What are the slow SQL queries?" "How many databases are there under the tenant?" "Is the backup normal?" "Are there any abnormal SQL queries?" "Which IPs are in the allowlist?" "PrivateLink connection information," "Which tenants can be restored?" "What is the node IP address?" and so on.
Covered scenarios: Cluster query, node information, tenant management, allowlist query, PrivateLink, database/user query, alert rules and history, backup status and recovery, TopSQL/SlowSQL/suspicious SQL diagnostics, SQL Outline binding, session statistics, monitoring metrics, organization and project permissions, etc.
odc-api
Description: The security specifications and ticket RESTful API documentation for ODC (OceanBase Developer Center). Use this to query resources such as ODC tickets (approval flow), security specifications (SQL check/window specification), risk levels, environment configurations, user roles, and operation audits.
Trigger scenarios: View tickets, My Tickets, Ticket Details, Ticket URL, Ticket Status, Pending Approval Tickets, SQL Check Specification, SQL Window Specification, Production Environment Rules, Security Specifications, Risk Levels, Approval Flow, Operation Audit, Who Changed What, etc.
Important note: When returning ticket information, each ticket must include a "Ticket Details URL" field, in the format: https://${ODC_DOMAIN}/index.html#/task?taskId={id}&taskType={type}&organizationId={orgId}.
powerrag-retrieval
Description: Achieves OceanBase documentation knowledge retrieval by calling the Workflow application API /service-api/v1/workflows/run. Suitable for retrieval and question-answering scenarios based on app-key (app- prefix).
Trigger scenario: Prioritize using this when inquiring about factual information about OceanBase/OB Cloud products.
Self-Improving + Proactive Agent
Description: Capable of self-reflection, self-criticism, self-learning, and self-organizing memory. The agent evaluates its own work, identifies errors, and continuously improves.
Trigger scenarios:
- When commands, tools, APIs, or operations fail.
- When a user corrects or rejects the agent's output.
- When the agent detects outdated or incorrect knowledge.
- When the agent discovers a better way to handle a task.
- When the user explicitly installs or references this skill for the current task.