Missions

更新时间:
复制 MD 格式

A Mission is an asynchronous, long-running human-machine collaboration mode in STAROps. You describe an O&M goal in natural language, and the AI autonomously plans, executes, validates, and reports the results over extended periods.

Overview

Complex O&M workflows often span hours, days, or even longer. They involve recurring inspections, multi-step remediation procedures, and cross-system coordination that are difficult to manage manually at scale. Missions address this by combining AI-driven autonomous execution with structured human oversight.

With Missions, a Digital Employee (Agent) autonomously handles task scheduling, tool invocation, result validation, and report generation. A built-in Human-in-the-Loop (HIL) mechanism pauses execution before high-risk operations and requests your confirmation, striking the right balance between automation and safety.

Key concepts

Concept

Description

Mission

A long-running O&M workflow that contains one or more Tasks. You define the high-level goal, and the AI generates a Blueprint to plan execution.

Task

An individual execution unit within a Mission. Each Task can be triggered on a schedule (cron), by an event, or manually.

Blueprint

The execution plan generated by the AI after intent alignment. A Blueprint defines the Tasks, their schedules, dependencies, and expected outputs.

Human-in-the-Loop (HIL)

A safety mechanism that pauses AI execution at critical decision points and requests your approval before proceeding.

Artifacts

Output files and reports generated during Task execution, including Task Reports, Mission Reports, and process snapshots.

How Missions work

A typical Mission follows this lifecycle:

  1. Intent alignment: You describe your O&M goal in natural language. The AI asks clarifying questions and confirms its understanding.

  2. Blueprint generation: The AI generates an execution plan (Blueprint) that breaks your goal into structured Tasks with schedules and dependencies.

  3. Plan review and launch: You review the Blueprint, make adjustments if needed, and launch the plan.

  4. Autonomous execution: The Digital Employee executes Tasks according to the schedule. For high-risk operations, HIL pauses execution and notifies you.

  5. Reporting and review: After each Task completes, the AI generates a Task Report. Periodic Mission Reports provide aggregated insights.

Task trigger methods

Tasks within a Mission support three trigger methods:

Trigger method

Description

Scheduled (cron)

The Task runs automatically on a recurring schedule. For example, run a Kubernetes cluster inspection every day at 02:00.

Event-driven

The Task runs automatically when a predefined condition is met. For example, start root cause analysis when an alert is triggered.

Manual

You trigger the Task on demand through a conversational command.

Topics in this section

  • Create and plan Missions: Learn how to align your intent with the AI, review the generated Blueprint, and launch the execution plan.

  • Task management and progress tracking: Understand Mission and Task states, manage the lifecycle, and track execution progress.

  • Task execution and human intervention: Monitor real-time execution output, respond to HIL requests, and add dynamic requirements.

  • Artifacts and report management: View Task Reports, Mission Reports, and process snapshots generated during execution.

  • Message center and notification configuration: Configure notification channels and manage in-app messages for Mission events.