This topic describes the core terms used throughout the STAROps (Intelligent O&M Platform) documentation. Understanding these concepts helps you get started with the platform and navigate the documentation effectively.
Glossary
The following table lists the core terms in STAROps. Each term is used consistently across all documentation.
Term | Definition |
Digital Employee (Agent) | The intelligent execution entity in STAROps. Each Digital Employee has a configurable name, role description, behavioral constraints (Default Rules), Skills, and permissions. A Digital Employee operates within assigned Data Scopes and can serve as a conversational counterpart in Intelligent Conversations or as the executor for Missions. |
Workspace | The runtime scope of a Digital Employee. A Workspace defines the observable resources, data sources, and service boundaries that the Digital Employee can access and act upon. |
Data Scope | The runtime perception and execution scope of a Digital Employee, defining the accessible resource boundary. In an intelligent conversation, a Data Scope consists of a workspace (a CMS 2.0 workspace) and a project (a Simple Log Service project). A Data Scope can be empty, or you can manually select one at the top of the dialog box. A Digital Employee can be associated with multiple Data Scopes. |
Mission | A top-level resource object that represents a long-running O&M goal. A Mission contains a Blueprint (execution plan) and one or more Tasks. Missions support cron-based, event-driven, and manual scheduling. |
Task | A single execution unit within a Mission. Each Task is triggered by a schedule, an event, or a manual action. When a Task completes, it generates a Task Report as part of the Mission Artifacts. |
Artifacts | The output produced by Task and Mission execution, including Task Reports (per-execution summaries) and Mission Reports (aggregated periodic summaries). |
Human-in-the-Loop (HIL) | A safety mechanism where the AI pauses before executing high-risk operations and requests your confirmation. HIL helps you maintain control over critical actions while still benefiting from AI automation. |
Skill | A reusable instruction module assigned to a Digital Employee. Each Skill encapsulates domain expertise, workflows, and best practices into a standardized capability unit that the Digital Employee loads on demand. |
Blueprint | The execution plan of a Mission. During the intent alignment phase, the AI generates a Blueprint that outlines the planned Tasks, schedules, and expected outputs. You can review and modify the Blueprint before launching the Mission. |
Unified Observability Model (UModel) | A data model that describes system topology and entity relationships, such as the mapping between clusters, nodes, services, and pods. UModel provides contextual information that enables the AI to perform correlation analysis across multiple data sources. |
Model Context Protocol (MCP) | An open standard protocol that connects Digital Employees to external systems and data sources. Through MCP services, Digital Employees can securely access databases, APIs, and cloud services to execute queries and automated operations. |
Default Rules | The behavioral configuration of a Digital Employee, written in Markdown. Default Rules define the role, data source focus, analysis logic, and output requirements that guide the Digital Employee during conversations and Task execution. |
How concepts relate to each other
The following list shows the key hierarchical relationships between STAROps concepts:
Digital Employee (Agent): the core execution entity
Equipped with Skills that encapsulate domain-specific execution capabilities.
Connected to MCP tools to access external systems and data sources.
Mission: a top-level resource object, associated with one Digital Employee and one Data Scope
Defined by a Blueprint that specifies Task schedules, trigger conditions, and execution parameters.
Contains one or more Tasks, each representing a specific execution instance.
Task: generates Artifacts upon completion
Produces a Task Report that records the execution process and conclusions.
Task Reports from multiple executions are aggregated into a Mission Report.
Mission states
A Mission lifecycle includes the following states:
State | Description |
Draft | The Mission has been created but no Blueprint has been confirmed and launched. |
Active | The Blueprint has been confirmed and launched. Tasks can be triggered and scheduled normally. |
Paused | The Mission has been manually disabled. Pending Tasks are suspended from scheduling. The Mission can be re-enabled to return to Active. |
Deleted | The Mission has been permanently deleted. All data is removed and cannot be recovered. |
The Deleted state is irreversible. If you only want to stop Task execution temporarily, use the disable function to switch the Mission to Paused instead.
Task trigger methods
A Task supports the following trigger methods:
Trigger method | Description | Example |
Scheduled trigger | Automatically triggered according to a cron expression. | Run a cluster health check at 02:00 every day. |
Manual trigger | Triggered by entering a command in the Mission conversation. | Enter an execution instruction in the Mission dialog box. |
A Task lifecycle includes the following states:
State | Description |
Running | The Task is currently executing. |
Success | The Task has completed execution successfully. |
Cancelled | The Task was cancelled before completion. |
Failed | The Task encountered an error during execution, such as a timeout. |