Newrelic Cli Skills

Alert Management

Manage alert policies, conditions, and notification channels via CLI.

---

List Alert Policies

newrelic alerts policy list

Get a Policy

newrelic alerts policy get --policyId <ID>

Create a Policy

newrelic alerts policy create \
  --name "My App - Performance" \
  --incidentPreference "PER_CONDITION_AND_TARGET"

Incident preference options:

• PER_POLICY — one incident per policy breach
• PER_CONDITION — one incident per condition
• PER_CONDITION_AND_TARGET — most granular, one per condition+entity

---

Alert Conditions

List Conditions for a Policy

newrelic alerts conditions list --policyId <POLICY_ID>

Create an APM Metric Condition

newrelic alerts apmCondition create \
  --policyId <POLICY_ID> \
  --name "High Response Time" \
  --type "apm_app_metric" \
  --metric "response_time_web" \
  --conditionScope "application" \
  --violationCloseTimer 24 \
  --threshold 2.0 \
  --thresholdDuration 5 \
  --thresholdOccurrences "ALL"

Delete a Condition

newrelic alerts conditions delete --conditionId <ID>

---

NRQL Alert Conditions

newrelic alerts nrqlCondition static create \
  --policyId <POLICY_ID> \
  --name "Error Rate > 5%" \
  --query "SELECT percentage(count(*), WHERE error IS true) FROM Transaction WHERE appName='my-app'" \
  --threshold 5 \
  --thresholdDuration 5 \
  --thresholdOccurrences "ALL" \
  --violationTimeLimitSeconds 86400

---

Notification Channels

# List channels
newrelic alerts channel list

# Create email channel
newrelic alerts channel create \
  --name "On-Call Email" \
  --type email \
  --configuration '{"recipients": "team@example.com", "include_json_attachment": false}'

---

View Open Incidents

newrelic nrql query --accountId $NEW_RELIC_ACCOUNT_ID --query "
  SELECT *
  FROM NrAiIncident
  WHERE event = 'open'
  SINCE 24 hours ago
  LIMIT 20
"

---

Check Entity Alert Severity

newrelic entity search --name "my-app" --type APPLICATION --domain APM | \
  jq '.[] | {name, alertSeverity}'

Severity values: NOT_CONFIGURED, NOT_ALERTING, WARNING, CRITICAL

newrelic-cli-skills

An OpenClaw agent skill for monitoring, querying, and managing New Relic observability data via the newrelic CLI.

What It Does

• Performance triage — identify slow transactions, DB bottlenecks, and error spikes
• NRQL queries — run ad-hoc queries against your New Relic account from the terminal
• Deployment markers — record releases so you can correlate deploys with performance changes
• Alert management — create and manage alert policies, conditions, and channels, including condition deletion workflows
• Infrastructure monitoring — host CPU, memory, disk, and process metrics
• Agent diagnostics — validate agent config and connectivity

Requirements

• `newrelic` CLI installed
• NEW_RELIC_API_KEY — User key (starts with NRAK-)
• NEW_RELIC_ACCOUNT_ID — Numeric account ID

Install CLI

Use a package manager or install the binary manually from the official New Relic CLI releases page.

Option 1: Homebrew (macOS)

brew install newrelic-cli

Option 2: Manual install from official releases

1. Open https://github.com/newrelic/newrelic-cli/releases 2. Download the archive for your OS/architecture 3. Extract it 4. Move the newrelic binary into a directory on your PATH 5. Verify with:

newrelic --version

Setup

newrelic profile add \
  --profile default \
  --apiKey $NEW_RELIC_API_KEY \
  --accountId $NEW_RELIC_ACCOUNT_ID \
  --region US

newrelic profile default --profile default

Sub-skills

Scripts

License

MIT

NRQL Patterns Reference

Transaction Performance

-- Response time percentiles
SELECT percentile(duration, 50, 75, 95, 99)
FROM Transaction
WHERE appName = 'my-app'
SINCE 1 hour ago

-- Apdex score (T = 0.5s threshold)
SELECT apdex(duration, 0.5)
FROM Transaction
WHERE appName = 'my-app'
TIMESERIES 5 minutes
SINCE 1 hour ago

-- Compare this week vs last week
SELECT average(duration)
FROM Transaction
WHERE appName = 'my-app'
SINCE 1 week ago
COMPARE WITH 1 week ago

-- Request volume by hour of day
SELECT count(*)
FROM Transaction
WHERE appName = 'my-app'
FACET hourOf(timestamp)
SINCE 1 week ago

Error Analysis

-- Error rate trend
SELECT percentage(count(*), WHERE error IS true)
FROM Transaction
WHERE appName = 'my-app'
TIMESERIES 5 minutes
SINCE 3 hours ago

-- Top errors by frequency
SELECT count(*)
FROM TransactionError
WHERE appName = 'my-app'
FACET error.class, message
SINCE 1 hour ago
LIMIT 20

-- Errors in the last 10 minutes
SELECT timestamp, transactionName, error.class, message
FROM TransactionError
WHERE appName = 'my-app'
SINCE 10 minutes ago
LIMIT 25

-- HTTP 5xx errors (if using HTTP status tracking)
SELECT count(*)
FROM Transaction
WHERE appName = 'my-app' AND response.status >= '500'
FACET response.status, name
SINCE 1 hour ago

Database Performance

-- DB time breakdown by operation type
SELECT average(duration)
FROM DatabaseTrace
WHERE appName = 'my-app'
FACET databaseVendor, category
SINCE 1 hour ago

-- Queries with high execution count (potential N+1)
SELECT count(*), average(duration)
FROM DatabaseTrace
WHERE appName = 'my-app'
FACET statement
SINCE 30 minutes ago
LIMIT 20
ORDER BY count(*) DESC

-- DB connection pool issues
SELECT count(*)
FROM TransactionError
WHERE appName = 'my-app' AND error.class LIKE '%connection%'
SINCE 1 hour ago

Infrastructure

-- All hosts CPU over time
SELECT average(cpuPercent)
FROM SystemSample
FACET hostname
TIMESERIES 10 minutes
SINCE 3 hours ago

-- Memory pressure (used > 85%)
SELECT latest(memoryUsedPercent)
FROM SystemSample
FACET hostname
WHERE memoryUsedPercent > 85
SINCE 5 minutes ago

-- Disk almost full (> 90%)
SELECT latest(diskUsedPercent), latest(mountPoint)
FROM StorageSample
FACET hostname
WHERE diskUsedPercent > 90
SINCE 10 minutes ago

Browser / Front-End

-- Page load time by page
SELECT average(duration)
FROM PageView
FACET pageUrl
SINCE 1 hour ago
LIMIT 20
ORDER BY average(duration) DESC

-- Core Web Vitals: LCP
SELECT average(largestContentfulPaint)
FROM PageViewTiming
FACET pageUrl
SINCE 1 hour ago

-- JS errors by page
SELECT count(*)
FROM JavaScriptError
FACET pageUrl, errorMessage
SINCE 1 hour ago
LIMIT 20

Deployments

-- Recent deployments
SELECT *
FROM Deployment
SINCE 1 week ago
LIMIT 20

-- Performance before vs after a specific deployment
SELECT average(duration)
FROM Transaction
WHERE appName = 'my-app'
SINCE '2026-02-25 10:00:00' UNTIL '2026-02-25 14:00:00'
TIMESERIES 5 minutes

Logs

-- Recent error logs
SELECT message, timestamp
FROM Log
WHERE level = 'ERROR'
SINCE 30 minutes ago
LIMIT 25

-- Log volume by level
SELECT count(*)
FROM Log
FACET level
SINCE 1 hour ago

Performance Triage Guide

Step-by-step guide for investigating a reported performance issue.

---

Step 1: Confirm the Problem

# Is there actually a performance issue right now?
./scripts/check-performance.sh "" 30

Look for:

• Avg response time > 1s (warning) or > 2s (critical)
• Error rate > 1% (warning) or > 5% (critical)
• RPM significantly lower than baseline (could indicate traffic drop or outage)

---

Step 2: Narrow to One App

# Which app is affected?
newrelic nrql query --accountId $NEW_RELIC_ACCOUNT_ID --query "
  SELECT average(duration) AS 'Avg (s)', percentage(count(*), WHERE error IS true) AS 'Error %'
  FROM Transaction
  FACET appName
  SINCE 30 minutes ago
  ORDER BY average(duration) DESC
"

---

Step 3: Find the Slow Endpoint

./scripts/top-slow-transactions.sh "my-app" 30

Identify the top offender by avg duration.

---

Step 4: Is It the Database?

newrelic nrql query --accountId $NEW_RELIC_ACCOUNT_ID --query "
  SELECT average(duration) AS 'Total', average(databaseDuration) AS 'DB',
         percentage(average(databaseDuration), average(duration)) AS 'DB %'
  FROM Transaction
  WHERE appName = 'my-app'
  FACET name
  SINCE 30 minutes ago
  ORDER BY average(databaseDuration) DESC
  LIMIT 10
"

If DB% > 60%: Focus on database query optimization. If DB% < 20% but total is slow: Look at external calls or application logic.

---

Step 5: Is It the Host?

# Check if host is under pressure
newrelic nrql query --accountId $NEW_RELIC_ACCOUNT_ID --query "
  SELECT average(cpuPercent), average(memoryUsedPercent)
  FROM SystemSample
  FACET hostname
  SINCE 30 minutes ago
"

CPU > 80%: Server is overloaded — check for runaway processes. Memory > 90%: Memory pressure — check for leaks or insufficient allocation.

---

Step 6: Is It a New Deployment?

newrelic apm deployment list --applicationId <APP_ID>

If a deploy happened in the last few hours, compare performance before/after:

newrelic nrql query --accountId $NEW_RELIC_ACCOUNT_ID --query "
  SELECT average(duration)
  FROM Transaction
  WHERE appName = 'my-app'
  TIMESERIES 10 minutes
  SINCE 3 hours ago
"

Look for an inflection point matching the deployment timestamp.

---

Step 7: Check for Errors

./scripts/error-report.sh "my-app" 30

A spike in errors often precedes or accompanies a performance degradation.

---

Step 8: External Dependencies

newrelic nrql query --accountId $NEW_RELIC_ACCOUNT_ID --query "
  SELECT average(duration), count(*)
  FROM ExternalTrace
  WHERE appName = 'my-app'
  FACET host
  SINCE 30 minutes ago
  ORDER BY average(duration) DESC
  LIMIT 10
"

If an external service (API, auth provider, CDN) is slow, that's often the root cause.

---

Decision Tree Summary

Slow app reported
    │
    ├─ Check all apps ──► Which one is slow?
    │
    ├─ Check transactions ──► Which endpoint?
    │
    ├─ DB % high? ──► Yes ──► Slow query / N+1 / missing index
    │                └─ No ──► External call? / CPU pressure? / code logic?
    │
    ├─ Host CPU/memory high? ──► Yes ──► Scale up or find runaway process
    │
    ├─ Recent deployment? ──► Yes ──► Regression — compare before/after
    │
    └─ Error spike? ──► Yes ──► Error-report.sh ──► Stack trace ──► Fix

---

What the CLI Cannot Do

• View distributed trace waterfalls (requires NR UI)
• Show flame graphs / code-level profiling (requires NR UI)
• Access custom dashboards interactively
• Resolve alerts or acknowledge incidents (UI only)

For deep dives requiring call graphs, hand the NRQL findings to the UI: https://one.newrelic.com/data-exploration → paste NRQL query