Application Insights Skill

Manage and analyze Application Insights using az monitor app-insights commands.

Discovery-First Rule

ALWAYS discover before acting. Never assume Application Insights resource names, instrumentation keys, or connection strings.

# Discover Application Insights resources
az monitor app-insights component show --output json \
  --query "[].{name:name, rg:resourceGroup, appId:appId, instrumentationKey:instrumentationKey, ingestionMode:ingestionMode, retentionDays:retentionInDays, kind:kind}"

Parallel Execution Requirement

ALL independent operations MUST run in parallel using background jobs (&) and wait.

for ai in $(echo "$components" | jq -c '.[]'); do
  {
    name=$(echo "$ai" | jq -r '.name')
    rg=$(echo "$ai" | jq -r '.rg')
    az monitor app-insights component show --app "$name" --resource-group "$rg" --output json
  } &
done
wait

Helper Functions

# Run App Insights query
query_app_insights() {
  local app="$1" query="$2" timespan="${3:-PT1H}"
  az monitor app-insights query --app "$app" --analytics-query "$query" --offset "$timespan" --output json
}

# Get request performance
get_request_perf() {
  local app="$1" timespan="${2:-PT1H}"
  query_app_insights "$app" "requests | where timestamp > ago(1h) | summarize totalCount=count(), failedCount=countif(success==false), avgDuration=avg(duration), p95Duration=percentile(duration,95), p99Duration=percentile(duration,99) by bin(timestamp, 5m) | sort by timestamp desc" "$timespan"
}

# Get dependency performance
get_dependency_perf() {
  local app="$1"
  query_app_insights "$app" "dependencies | where timestamp > ago(1h) | summarize totalCount=count(), failedCount=countif(success==false), avgDuration=avg(duration) by target, type | sort by totalCount desc | take 20"
}

# Get exception summary
get_exceptions() {
  local app="$1"
  query_app_insights "$app" "exceptions | where timestamp > ago(1h) | summarize count() by type, outerMessage | sort by count_ desc | take 20"
}

Common Operations

1. Application Health Overview

app_id=$(az monitor app-insights component show --app "$APP" --resource-group "$RG" --query "appId" -o tsv)

# Request rate and failure rate
query_app_insights "$app_id" "
  requests
  | where timestamp > ago(1h)
  | summarize
      TotalRequests=count(),
      FailedRequests=countif(success==false),
      FailureRate=round(100.0*countif(success==false)/count(), 2),
      AvgDuration=round(avg(duration), 2),
      P95=round(percentile(duration, 95), 2)
  by bin(timestamp, 5m)
  | sort by timestamp desc"

2. Dependency Tracking

# Slowest dependencies
query_app_insights "$app_id" "
  dependencies
  | where timestamp > ago(1h)
  | summarize
      calls=count(),
      failures=countif(success==false),
      avgMs=round(avg(duration), 2),
      p95Ms=round(percentile(duration, 95), 2)
  by target, type, name
  | sort by avgMs desc
  | take 15"

# Failed dependency calls
query_app_insights "$app_id" "
  dependencies
  | where timestamp > ago(1h) and success == false
  | summarize count() by target, type, resultCode
  | sort by count_ desc"

3. Availability Test Results

# List availability tests
az monitor app-insights web-test list --resource-group "$RG" --output json \
  --query "[].{name:name, enabled:enabled, kind:kind, frequency:frequency, timeout:timeout, locations:locations[].Id}"

# Check availability results
query_app_insights "$app_id" "
  availabilityResults
  | where timestamp > ago(24h)
  | summarize
      successRate=round(100.0*countif(success==1)/count(), 2),
      avgDuration=round(avg(duration), 2),
      totalTests=count()
  by name
  | sort by successRate asc"

4. Smart Detection and Anomalies

# Check smart detection alerts
az monitor app-insights component show --app "$APP" --resource-group "$RG" --output json \
  --query "{smartDetection:smartDetection}"

# Query for anomalies in failure rate
query_app_insights "$app_id" "
  requests
  | where timestamp > ago(24h)
  | summarize failRate=round(100.0*countif(success==false)/count(), 2), total=count() by bin(timestamp, 1h)
  | sort by timestamp desc"

5. End-to-End Transaction Diagnostics

# Trace a specific operation
query_app_insights "$app_id" "
  union requests, dependencies, exceptions, traces
  | where operation_Id == '$OPERATION_ID'
  | sort by timestamp asc
  | project timestamp, itemType, name, duration, success, resultCode, message, type"

# Slowest operations
query_app_insights "$app_id" "
  requests
  | where timestamp > ago(1h)
  | summarize avgDuration=avg(duration), p95=percentile(duration, 95), count=count() by name
  | where count > 10
  | sort by p95 desc
  | take 10"

Output Format

Present results as a structured report:

Azure App Insights Report
═════════════════════════
Resources discovered: [count]

Resource       Status    Key Metric    Issues
──────────────────────────────────────────────
[name]         [ok/warn] [value]       [findings]

Summary: [total] resources | [ok] healthy | [warn] warnings | [crit] critical
Action Items: [list of prioritized findings]

Target ≤50 lines of output. Use tables for multi-resource comparisons.

Anti-Hallucination Rules

1. NEVER assume resource names — always discover via CLI/API in Phase 1 before referencing in Phase 2.
2. NEVER fabricate metric names or dimensions — verify against the service documentation or --help output.
3. NEVER mix CLI commands between service versions — confirm which version/API you are targeting.
4. ALWAYS use the discovery → verify → analyze chain — every resource referenced must have been discovered first.
5. ALWAYS handle empty results gracefully — an empty response is valid data, not an error to retry.

Counter-Rationalizations

Shortcut	Counter	Why
"I'll skip discovery and check known resources"	Always run Phase 1 discovery first	Resource names change, new resources appear — assumed names cause errors
"The user only asked for a quick check"	Follow the full discovery → analysis flow	Quick checks miss critical issues; structured analysis catches silent failures
"Default configuration is probably fine"	Audit configuration explicitly	Defaults often leave logging, security, and optimization features disabled
"Metrics aren't needed for this"	Always check relevant metrics when available	API/CLI responses show current state; metrics reveal trends and intermittent issues
"I don't have access to that"	Try the command and report the actual error	Assumed permission failures prevent useful investigation; actual errors are informative

Common Pitfalls

1. Sampling: Application Insights may sample data at high volumes. Check itemCount field to see if counts are projected, not exact.
2. Ingestion delay: Data can take 2-5 minutes to appear. Do not query for the last few minutes expecting real-time results.
3. Workspace-based vs classic: Workspace-based App Insights stores data in Log Analytics. Use az monitor log-analytics query for cross-resource queries.
4. Duration units: duration in KQL is in milliseconds for requests/dependencies. Do not confuse with seconds.
5. Connection string vs instrumentation key: Instrumentation keys are deprecated for new resources. Use connection strings for SDK configuration checks.