Reverse-DNS namespace prefix

No bare custom attribute names

http.route in metric dimensions

No url.path in metrics

No user/order IDs in metric dimensions

User identity on spans not resource

service.name on resource

deployment.environment.name used

Registry attributes not reinvented

High-cardinality values on spans

service.name as resource attr

service.version as resource attr

deployment.environment.name

k8s.pod.uid not k8s.pod.ip

service.name NOT on spans

k8s metadata NOT on spans

peer.service or server.address on client spans

Correct span kinds

No manual dash0.* attributes

service.instance.id set

UnixNano(Now()) for stale drop

Stale drop uses filter processor

Timestamp backfill observed_time

Timestamp backfill log.time

Enum constant STATUS_CODE_ERROR

IsMatch() for pattern filtering

filter processor for metric drops

nil for absence checks

error_mode set explicitly

error_mode ignore in production

where clause nil guards on attrs

Manual SERVER root span

CLIENT kind for outbound calls

AlwaysOn sampler preserved

ERROR only after retries exhausted

Retry attempts as span events

ERROR status message

No span.recordException

Batch span instead of per-item spans

Exception log trace context

forceFlush on shutdown

No more than 10 INTERNAL spans

Enrichment on all pipelines

k8s.pod.uid first in pod_association

Workload name and UID extracted

resourcedetection override false

resource processor for cluster attributes

Processor ordering: enrichment after memory_limiter

Processor ordering: resource after enrichment

k8sattributes passthrough: false

memory_limiter first

k8s.cluster.uid kube-system reference

No batch processor

http.request.method used

http.response.status_code used

url.full on client spans

url.path on server spans

url.full NOT on server spans

server.address used

url.scheme used

user_agent.original used

Metric name updated

Duration unit seconds

_OTHER normalization

Histogram for durations

Counter for monotonic counts

UpDownCounter for queue depth

UCUM units specified

No unit in metric name

No semconv namespace clash

No high-cardinality metric attributes

Bounded attribute values only

No duplicate auto-instrumented metric

Metric shape test

Consistent units across instances

Metric name not an attribute key

Correct SDK activation flag

OTEL_TRACES_EXPORTER set

gRPC protocol configured

4xx SERVER span status

5xx SERVER span status

No span.recordException

Exception log trace context

URL query param redaction

No PII or credentials in telemetry

Structured logging only

Trace-log correlation in handlers

forceFlush on crash handlers

memory_limiter first

No batch processor

sending_queue with file_storage

No hardcoded token

gzip compression

No debug exporter

Separate signal pipelines

health_check extension

0.0.0.0 bind address

memory_limiter sizing

file_storage extension declared

All declared components used

nil guard on redacted attrs

nil keyword not null

error_mode set explicitly

error_mode ignore in production

SHA256 hash for emails/IDs

replace_pattern for partial masking

filter processor for full drops

redaction after enrichment

delete_key for never-export attrs

set REDACTED for auth headers

debug exporter for validation

signaltometrics not spanmetrics

exponential_histogram used

Connector before sampling in pipeline

Metrics pipeline uses connector as receiver

Semconv metric names

Semconv conditions used

service.name in resource attributes

AlwaysOn sampler recommendation

Tail sampling policies include errors and latency

memory_limiter first

No batch processor

UUID pattern normalization

Numeric ID normalization

nil guard on url.path

nil guard on http.route

limit() for attribute count

truncate_all() for value length

resource processor for static attrs

transform to copy resource to span

error_mode set explicitly

error_mode ignore production

nil keyword not null