---

name: metrics description: This skill should be used when the user asks about resource usage, CPU, memory, network, disk, or service performance. Covers questions like "how much memory is my service using" or "is my service slow". allowed-tools: Bash(railway:*)

Service Metrics

Query resource usage metrics for Railway services.

When to Use

• User asks "how much memory is my service using?"
• User asks about CPU usage, network traffic, disk usage
• User wants to debug performance issues
• User asks "is my service healthy?" (combine with service skill)

Prerequisites

Get environmentId and serviceId from linked project:

Copyrailway status --json

Extract:

• environment.id → environmentId
• service.id → serviceId (optional - omit to get all services)

MetricMeasurement Values

MetricTag Values (for groupBy)

Query

Copyquery metrics(
  $environmentId: String!
  $serviceId: String
  $startDate: DateTime!
  $endDate: DateTime
  $sampleRateSeconds: Int
  $averagingWindowSeconds: Int
  $groupBy: [MetricTag!]
  $measurements: [MetricMeasurement!]!
) {
  metrics(
    environmentId: $environmentId
    serviceId: $serviceId
    startDate: $startDate
    endDate: $endDate
    sampleRateSeconds: $sampleRateSeconds
    averagingWindowSeconds: $averagingWindowSeconds
    groupBy: $groupBy
    measurements: $measurements
  ) {
    measurement
    tags {
      deploymentInstanceId
      deploymentId
      serviceId
      region
    }
    values {
      ts
      value
    }
  }
}

Example: Last Hour CPU and Memory

Use heredoc to avoid shell escaping issues:

Copybash <<'SCRIPT'
START_DATE=$(date -u -v-1H +"%Y-%m-%dT%H:%M:%SZ" 2>/dev/null || date -u -d "1 hour ago" +"%Y-%m-%dT%H:%M:%SZ")
ENV_ID="your-environment-id"
SERVICE_ID="your-service-id"

VARS=$(jq -n \
  --arg env "$ENV_ID" \
  --arg svc "$SERVICE_ID" \
  --arg start "$START_DATE" \
  '{environmentId: $env, serviceId: $svc, startDate: $start, measurements: ["CPU_USAGE", "MEMORY_USAGE_GB"]}')

scripts/railway-api.sh \
  'query metrics($environmentId: String!, $serviceId: String, $startDate: DateTime!, $measurements: [MetricMeasurement!]!) {
    metrics(environmentId: $environmentId, serviceId: $serviceId, startDate: $startDate, measurements: $measurements) {
      measurement
      tags { deploymentId region serviceId }
      values { ts value }
    }
  }' \
  "$VARS"
SCRIPT

Example: All Services in Environment

Omit serviceId and use groupBy to get metrics for all services:

Copybash <<'SCRIPT'
START_DATE=$(date -u -v-1H +"%Y-%m-%dT%H:%M:%SZ" 2>/dev/null || date -u -d "1 hour ago" +"%Y-%m-%dT%H:%M:%SZ")
ENV_ID="your-environment-id"

VARS=$(jq -n \
  --arg env "$ENV_ID" \
  --arg start "$START_DATE" \
  '{environmentId: $env, startDate: $start, measurements: ["CPU_USAGE", "MEMORY_USAGE_GB"], groupBy: ["SERVICE_ID"]}')

scripts/railway-api.sh \
  'query metrics($environmentId: String!, $startDate: DateTime!, $measurements: [MetricMeasurement!]!, $groupBy: [MetricTag!]) {
    metrics(environmentId: $environmentId, startDate: $startDate, measurements: $measurements, groupBy: $groupBy) {
      measurement
      tags { serviceId region }
      values { ts value }
    }
  }' \
  "$VARS"
SCRIPT

Time Parameters

Tip: For last hour, calculate startDate as now - 1 hour in ISO format.

Output Interpretation

Copy{
  "data": {
    "metrics": [
      {
        "measurement": "CPU_USAGE",
        "tags": { "deploymentId": "...", "serviceId": "...", "region": "us-west1" },
        "values": [
          { "ts": "2024-01-01T00:00:00Z", "value": 0.25 },
          { "ts": "2024-01-01T00:01:00Z", "value": 0.30 }
        ]
      }
    ]
  }
}

• ts - timestamp in ISO format
• value - metric value (cores for CPU, GB for memory/disk/network)

Composability

• Get IDs: Use status skill or railway status --json
• Check service health: Use service skill for deployment status
• View logs: Use deployment skill if metrics show issues
• Scale service: Use environment skill to adjust resources

Error Handling

Empty/Null Metrics

Services without active deployments return empty metrics arrays. When processing with jq, handle nulls:

Copy# Safe iteration - skip nulls
jq -r '.data.metrics[]? | select(.values != null and (.values | length) > 0) | ...'

# Check if metrics exist before processing
jq -e '.data.metrics | length > 0' response.json && echo "has metrics"

No Metrics Data

Service may be new or have no traffic. Check:

• Service has active deployment (stopped services have no metrics)
• Time range includes deployment period

Invalid Service/Environment ID

Verify IDs with railway status --json.

Permission Denied

User needs access to the project to query metrics.