alert metrics CLI doesn't support detector/workflow log alerts and list/view are inconsistent

[sergical]

Summary

sentry alert metrics currently leads users/agents into legacy metric alert-rule behavior and does not support the new detector/workflow model required for log/EAP alerts.

In a Next.js project (demo/ai-engineer-conf) I tried to create log-based operational alerts for structured logs (trace_item_log). The CLI exposed sentry alert metrics create/list/view, but the behavior was confusing and partially broken for this use case.

Repro / observed behavior

1. create rejects logs as a dataset

sentry alert metrics create demo \
  --name 'AI chat failures' \
  --project ai-engineer-conf \
  --dataset logs \
  --query 'action:ai.chat (result:error OR result:stream_error)' \
  --aggregate 'count()' \
  --time-window 5 \
  --trigger '{"alertThreshold":1,"resolveThreshold":0,"label":"critical","actions":[{"id":"sentry.mail.actions.NotifyEmailAction","targetType":"Team","targetIdentifier":"2279506"}]}'

Returns:

Error: dataset must be one of: errors, transactions, sessions, events, spans, metrics.

However, log alerts are supported via the new monitors/alerts API using:

{
  "dataset": "events_analytics_platform",
  "eventTypes": ["trace_item_log"]
}

2. create trigger shape differs from API shape

The CLI help examples use legacy action ids like:

{"id":"sentry.mail.actions.NotifyEmailAction","targetType":"Team","targetIdentifier":1}

When posting directly to /api/0/organizations/{org}/alert-rules/, the API expects metric alert trigger actions shaped like:

{"type":"email","targetType":"team","targetIdentifier":"2279506"}

3. list does not list log/EAP alert rules that view can view

After creating a legacy EAP/log alert rule via direct API:

sentry alert metrics view demo/441020 --json

worked and returned the alert rule, including:

{
  "id": "441020",
  "dataset": "events_analytics_platform",
  "eventTypes": ["trace_item_log"]
}

But:

sentry alert metrics list demo/ --json --fresh

returned:

{"data":[],"hasMore":false,"hasPrev":false}

Direct API listing did return it:

sentry api '/api/0/organizations/demo/alert-rules/?project=4511626979573760'

4. The command appears to use legacy /alert-rules/, but we should be using detector/workflow

For the actual fix, I had to use the new model directly:

• POST /api/0/organizations/demo/projects/ai-engineer-conf/detectors/
• POST /api/0/organizations/demo/workflows/

Detector data source example that worked:

{
  "name": "AI chat failures",
  "type": "metric_issue",
  "dataSources": [
    {
      "aggregate": "count()",
      "dataset": "events_analytics_platform",
      "environment": null,
      "eventTypes": ["trace_item_log"],
      "query": "action:ai.chat (result:error OR result:stream_error)",
      "queryType": 1,
      "timeWindow": 300,
      "extrapolationMode": "unknown"
    }
  ],
  "config": {"detectionType": "static"},
  "conditionGroup": {
    "logicType": "any",
    "conditions": [
      {"type": "gt", "comparison": 0, "conditionResult": 75},
      {"type": "lte", "comparison": 0, "conditionResult": 0}
    ],
    "actions": []
  }
}

Workflow connection example that worked:

{
  "name": "AI Engineer operational alerts",
  "detectorIds": [7659187],
  "triggers": {
    "logicType": "any-short",
    "conditions": [
      {"type": "first_seen_event", "comparison": true, "conditionResult": true},
      {"type": "reappeared_event", "comparison": true, "conditionResult": true},
      {"type": "regression_event", "comparison": true, "conditionResult": true}
    ]
  },
  "actionFilters": [
    {
      "logicType": "all",
      "conditions": [
        {"type": "issue_priority_greater_or_equal", "comparison": 75, "conditionResult": true}
      ],
      "actions": [
        {
          "type": "email",
          "integrationId": null,
          "data": {},
          "config": {
            "targetType": "team",
            "targetDisplay": null,
            "targetIdentifier": "2279506"
          },
          "status": "active"
        }
      ]
    }
  ]
}

Expected behavior

Either:

1. sentry alert metrics should support the new detector/workflow-backed alert model for log/EAP alerts, including logs or events_analytics_platform + trace_item_log, and list/view should be consistent.

or

2. The command should clearly state that it manages legacy alert rules only, and the CLI should expose separate commands for new monitors/detectors/workflows.

Environment

sentry --version
# 0.39.0-dev.1782404180