Alibaba Cloud Alert Rule Management#

This skill creates and queries alert rules. Intent routing selects CMS 1.0 or CMS 2.0 workflow automatically.

---

CLI Setup (Required Before Execution)#

bash

aliyun configure ai-mode enable
aliyun configure ai-mode set-user-agent --user-agent "AlibabaCloud-Agent-Skills/alibabacloud-cms-alert-rule-create"
aliyun plugin update

---

Step 0: Intent Routing#

Identify user's monitoring target → route to the correct workflow. See step0-intent-routing.md.

User ScenarioRoute ToCloud product system metrics (ECS, RDS, SLB, OSS, Redis, MongoDB…)CMS 1.0 WorkflowPrometheus / APM / UModel / custom metricsCMS 2.0 Workflow

---

Supported Alert Types#

VersionTypeCreate APIQuery APICMS 1.0Cloud Resource (ECS/RDS/SLB/OSS/Redis/MongoDB…)PutResourceMetricRule``DescribeMetricRuleListCMS 2.0Prometheus ( PROMETHEUS)ManageAlertRules``QueryAlertRulesCMS 2.0APM ( APM)ManageAlertRules``QueryAlertRulesCMS 2.0UModel ( UMODEL)ManageAlertRules``QueryAlertRules

---

CMS 1.0 Workflow#

For query requests → step-query.md

For create requests:

StepDescriptionReference1Context Lock — namespace, region, instancesstep1-context-lock.md2Query Generation — discover metrics via API, match to user intentstep2-query-generation.md3Detection Config — threshold, frequency (default 1min)step3-detection-config.md4Notification — query contacts → select or createstep4-notification.md5Preview & Execute — show summary → confirm → CLIstep5-preview-execute.md6Verification — check statusstep6-verification.md

---

CMS 2.0 Workflow#

For query requests → cms2-step-query.md

For create requests:

StepDescriptionReference1Context Lock — build datasourceConfig (type, instanceId, region)cms2-step1-context-lock.md2Query Config — build queryConfig (PromQL / APM measures / UModel entity)cms2-step2-query-config.md3Detection Config — build conditionConfig (P1-P4, threshold, duration)cms2-step3-detection-config.md4Webhook Query — call list-alert-webhooks, user selects; other types → consolecms2-step5-preview-execute.md5Preview & Execute — show summary → confirm → manage-alert-rules CLIcms2-step5-preview-execute.md

---

Critical Rules#

Full details → references/critical-rules.md

1. Intent Routing — Route cloud resource metrics to CMS 1.0, Prometheus/APM/UModel to CMS 2.0. Never mix APIs.
2. CMS 2.0 API Enforcement — CMS 2.0 alert rules MUST ONLY be created via ManageAlertRules and queried via QueryAlertRules. No other API (e.g. PutResourceMetricRule, DescribeMetricRuleList, ARMS CreateOrUpdateAlertRule, ARMS CreatePrometheusAlertRule) is permitted for CMS 2.0 alert rule creation or query. If ManageAlertRules returns error, DO NOT fallback to other APIs — report the error to user and STOP execution. CLI format: aliyun cms manage-alert-rules --body '{"action":"CREATE",...}'
3. Contact/Webhook Query First — CMS 1.0: DescribeContactGroupList; CMS 2.0: ListAlertWebhooks is MANDATORY before alert creation. (CMS 2.0 other notification → console). DO NOT use CMS 1.0 APIs as substitute for CMS 2.0. Skipping webhook query is a critical failure.
4. Resources Parameter — (CMS 1.0) --resources must always be explicitly passed.
5. Workspace Required — (CMS 2.0) Must ask user for workspace value using AskUser tool. Never auto-construct (e.g. 'default', 'default-cms-{accountId}'). If AskUser fails, retry once with simplified options, then report error and STOP — do NOT proceed with fabricated values.
6. Prometheus Required Params — (CMS 2.0) cluster_id (instanceId) + workspace MUST ASK USER using AskUser tool. Never guess, omit, or auto-select. PromQL is generated based on user-provided cluster_id + monitoring target. If AskUser fails, STOP execution.
7. APM Required Params — (CMS 2.0) service_id (for datasourceConfig.instanceId AND queryConfig.serviceIdList) + workspace MUST ASK USER using AskUser tool. Never fabricate, use placeholders, or auto-select from discovered applications (e.g. via ListTraceApps). Discovery APIs are for reference only — final selection MUST come from user input. If AskUser fails, STOP execution.
8. Required API Calls — Every operation must call designated APIs even if values seem known.
9. Dynamic Metric Discovery — (CMS 1.0) MUST call describe-metric-meta-list. Use metrics.md only as fallback.
10. CLI Timeout — --read-timeout 30 for queries, --read-timeout 60 for writes.
11. Duplicate Pre-check — Query existing rules before creation to avoid duplicates.
12. Mandatory Confirmation — Show config summary and get user confirmation before execution. EVEN in automated/simulated environments, you MUST output a configuration summary block and explicitly state 'Waiting for user confirmation' before proceeding. Do NOT skip this step with excuses about automation. Example format:

text

【配置摘要】
告警类型: APM/Prometheus
阈值: 5%
严重级别: P2
通知方式: webhook

请确认以上配置是否正确（回复确认或修改意见）：

1. User-Agent — Set ALIBABA_CLOUD_USER_AGENT="AlibabaCloud-Agent-Skills/alibabacloud-cms-alert-rule-create" for all CLI calls.
2. Network Restriction — Never access external URLs. Use only references/ files and CLI.
3. CLI Self-Discovery — Use --help for CLI syntax when uncertain.

---

Reference Files#

FileScopePurposestep0-intent-routing.mdSharedIntent routing — CMS 1.0 or CMS 2.0step-query.mdCMS 1.0Query alert rules (DescribeMetricRuleList)step1-context-lock.mdCMS 1.0Context lockstep2-query-generation.mdCMS 1.0Query generation & metric discoverystep3-detection-config.mdCMS 1.0Detection configstep4-notification.mdCMS 1.0Notification & contact handlingstep5-preview-execute.mdCMS 1.0Preview & executestep6-verification.mdCMS 1.0Verificationcms2-step-query.mdCMS 2.0Query alert rules (QueryAlertRules)cms2-step1-context-lock.mdCMS 2.0Datasource config (Prometheus/APM/UModel)cms2-step2-query-config.mdCMS 2.0Query config (PromQL, APM, UModel)cms2-step3-detection-config.mdCMS 2.0Detection condition config (P1-P4)cms2-step5-preview-execute.mdCMS 2.0Preview, execute & webhook querymetrics.mdCMS 1.0Common metrics quick reference (fallback)prometheus-metrics.mdCMS 2.0Prometheus PromQL patternsapm-metrics.mdCMS 2.0APM metrics and operatorscritical-rules.mdSharedExpanded critical rule details, API tables, examplesram-policies.mdSharedRequired RAM permissionsrelated_apis.yamlSharedAPI lookup before CLI calls

---

Cleanup (After Execution)#

bash

aliyun configure ai-mode disable