---
name: "alarm"
description: "Alarm operations skill. List alerts, analyze alert rule context with datasource-enriched resource facts, and perform alert status operations with remediation guidance."
provider_type: "smartcmp"
instance_required: "true"

# === LLM Context Fields ===
triggers:
  - list alarms
  - show alerts
  - analyze alert
  - mute alert
  - resolve alert
  - reopen alert
  - operate alert
  - 查看告警
  - 查看我的告警
  - 告警列表
  - 分析告警
  - 静默告警
  - 解决告警
  - 重新打开告警
  - 告警操作

use_when:
  - User wants to inspect alarms or alerts
  - User needs rule-aware, resource-aware alarm analysis with remediation recommendations
  - User wants to mute, resolve, or reopen alerts

avoid_when:
  - User wants approval actions (use approval skill)
  - User wants resource requests (use request skill)
  - User only wants standalone reference data browsing without alarm analysis (use datasource skill)

examples:
  - "Show current alarms"
  - "Analyze alert ALM-1001"
  - "Mute alert ALM-1001"
  - "Resolve alerts ALM-1001 and ALM-1002"
  - "查看我的告警"
  - "查看告警"
  - "分析告警 ALM-1001"
  - "静默告警 ALM-1001"

related:
  - approval
  - datasource

tool_list_name: "smartcmp_list_alerts"
tool_list_description: "List SmartCMP triggered alerts with optional filters for status, severity level, time range, deployment, entity instance, node instance, alarm type, alarm category, and keyword query."
tool_list_entrypoint: "scripts/list_alerts.py"
tool_list_groups:
  - cmp
  - alarm
tool_list_capability_class: "provider:smartcmp"
tool_list_priority: 100

tool_analyze_name: "smartcmp_analyze_alert"
tool_analyze_description: "Analyze one SmartCMP alert with rule context, datasource-enriched resource facts, assessment, and remediation guidance. Always pass a real SmartCMP alertId as alert_id. If the user refers to a numbered result from a prior alert list, resolve that display index from the previous smartcmp_list_alerts metadata and pass that item's alertId, not the display index."
tool_analyze_entrypoint: "scripts/analyze_alert.py"
tool_analyze_groups:
  - cmp
  - alarm
tool_analyze_capability_class: "provider:smartcmp"
tool_analyze_priority: 120
tool_analyze_cli_positional:
  - alert_id
tool_analyze_parameters: |
  {
    "type": "object",
    "properties": {
      "alert_id": {
        "type": "string",
        "description": "SmartCMP alertId to analyze. If the user references a numbered alert result, use the alertId from the matching previous smartcmp_list_alerts metadata item; do not pass the display index."
      },
      "days": {
        "type": "integer",
        "description": "Trend lookback window in days. Default: 7.",
        "default": 7
      }
    },
    "required": ["alert_id"]
  }

tool_operate_name: "smartcmp_operate_alert"
tool_operate_description: "Perform validated status operations (mute, resolve, reopen) on one or more SmartCMP alerts. Always pass real SmartCMP alert IDs; resolve numbered result references from prior smartcmp_list_alerts metadata before calling."
tool_operate_entrypoint: "scripts/operate_alert.py"
tool_operate_groups:
  - cmp
  - alarm
tool_operate_capability_class: "provider:smartcmp"
tool_operate_priority: 150
tool_operate_cli_positional:
  - alert_ids
tool_operate_cli_split:
  - alert_ids
tool_operate_parameters: |
  {
    "type": "object",
    "properties": {
      "alert_ids": {
        "type": "string",
        "description": "One or more SmartCMP alert IDs separated by spaces. If the user references numbered alert results, first resolve each display index to its alertId from prior smartcmp_list_alerts metadata."
      },
      "action": {
        "type": "string",
        "enum": ["mute", "resolve", "reopen"],
        "description": "Status operation to perform on the target alert IDs."
      }
    },
    "required": ["alert_ids", "action"]
  }
---

# alarm

Alarm workflow for triggered alert retrieval, rule-aware analysis, and
status operations.

## Purpose

Provide alarm-management capabilities:
- List triggered alerts with machine-readable metadata
- Analyze one alert with normalized facts, datasource-enriched resource context, assessment, and remediation guidance
- Perform validated status operations (`mute`, `resolve`, `reopen`)

## Scripts

| Script | Description | Location |
|--------|-------------|----------|
| `list_alerts.py` | Query `/alarm-alert` and emit `##ALARM_META_START##` metadata | `scripts/` |
| `analyze_alert.py` | Fetch alert + rule context, enrich related resources via datasource `list_resource.py`, and emit `##ALARM_ANALYSIS_START##` output | `scripts/` |
| `operate_alert.py` | Call `/alarm-alert/operation` for validated status changes | `scripts/` |

## Workflow

See [references/WORKFLOW.md](references/WORKFLOW.md) for the supported workflow.

## Resource Enrichment

During alert analysis, this skill should silently reuse the datasource skill's
shared `../datasource/scripts/list_resource.py` flow when the alert exposes
`entityInstanceId` or `nodeInstanceId`.

- Resolve related SmartCMP resources before finalizing the analysis narrative.
- Merge normalized `type + properties` resource facts into the alert analysis payload.
- If resource lookup is unavailable, continue the alert analysis with core alert
  and policy facts, and treat resource enrichment as best-effort context.