Instruments Java, Python, Node.js, .NET, and Go applications with OpenTelemetry SDKs to send traces, metrics, and logs to Coralogix. Use for SDK-side OTel setup, OTLP exporter env vars, Coralogix resource attributes, APM transaction samplers, Kubernetes Operator injection, or debugging missing traces, metrics, logs, or no telemetry from an application. Not for OTel Collector config (use opentelemetry-collector), OTTL authoring (use opentelemetry-ottl), eBPF instrumentation, or Lambda layers.
Installation
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Instruments Java, Python, Node.js, .NET, and Go applications with OpenTelemetry SDKs to send traces, metrics, and logs to Coralogix. Use for SDK-side OTel setup, OTLP exporter env vars, Coralogix resource attributes, APM transaction samplers, Kubernetes Operator injection, or debugging missing traces, metrics, logs, or no telemetry from an application. Not for OTel Collector config (use opentelemetry-collector), OTTL authoring (use opentelemetry-ottl), eBPF instrumentation, or Lambda layers.
license
Apache-2.0
metadata
{"version":"0.1.0","integration":"otel-sdk-instrumentation","signals":["logs","metrics","traces"],"deployment":["kubernetes","docker","aws"],"triggers":{"description":"Load when a user is adding or reviewing OpenTelemetry SDK instrumentation in an application — choosing a language, wiring up an exporter, setting env vars, or debugging missing telemetry from the SDK side (not from the collector side).\n","always":false,"file_patterns":["**/*instrumentation*.js","**/*instrumentation*.ts","**/*instrumentation*.py","**/*otel*.py","**/*otel*.java","**/*opentelemetry*.java","**/*opentelemetry*.cs","**/*opentelemetry*.go"]},"docs":"https://coralogix.com/docs/opentelemetry/"}
OpenTelemetry SDK Instrumentation for Coralogix
When to Use This Skill
Use case
What to do
Choose a language / decide where to start
Load language-selector.md first — it has the decision tree for language, signal, mode, and export path
Always call list_references() first, then read_reference(...) for the relevant files (language selector + the target language + endpoints; add troubleshooting only when debugging). Do not answer from memory if the references could affect correctness.
Use the structure/checklists in output-templates.md for any user-facing response; keep your own prose minimal and actionable.
Always include in generated answers:
Required env vars (names only) — include these in any “copy/paste” setup:
OTEL_EXPORTER_OTLP_ENDPOINT, OTEL_EXPORTER_OTLP_HEADERS, OTEL_SERVICE_NAME, OTEL_RESOURCE_ATTRIBUTES
(plus OTEL_*_EXPORTER / OTEL_EXPORTER_OTLP_*_PROTOCOL when choosing signals/protocol). For exact per-language endpoint formats and the Python %20 env-var rule, use coralogix-endpoints.md.
Required resource attributes — service.name (APM catalog/transactions), cx.application.name + cx.subsystem.name (routing, TCO, APM grouping), telemetry.sdk.language (auto-set; do not strip). Missing any silently degrades APM. Additional k8s/host attributes are covered in each language reference.
Telemetry quality. Low-cardinality span names, no sensitive data, bounded metric attributes, structured single-line logs with trace/span IDs, validation steps for the signal. Never approve user-scoped values (user_id, tenant_id, request_id, session_id) as metric labels — they cause unbounded cardinality and will explode metric storage.
Direct vs collector routing rule: Default to direct OTLP for dev/test environments, simple single-service production deployments, or when no collector infrastructure exists. Default to via-collector for Kubernetes workloads needing pod/namespace enrichment (k8sattributes), tail sampling, multi-service shared key management, or buffered retry/file persistence requirements. When the deployment context is unknown, show both options using the tradeoff table in language-selector.md.
Collector config is out of scope. Redirect to opentelemetry-collector — do not partially answer pipeline or processor questions.
High-Risk Gotchas (Do Not Guess)
Keep this list short and only use it when the user’s symptom matches. For details, defer to the language references.
Java/.NET gRPC endpoint: must be a full URI https://ingress.<region>.coralogix.com:443 (URI parsing breaks bare host:port).
Python auth header: env var needs %20 (Authorization=Bearer%20<KEY>); programmatic headers use a tuple sequence with lowercase key + literal space (("authorization", f"Bearer {token}")).
Node.js HTTP/proto: exporter url must include /v1/traces (https://ingress.<region>.coralogix.com:443/v1/traces); gRPC doesn’t.
Go gRPC:credentials.NewTLS(nil) and bare endpoint ingress.<region>.coralogix.com:443 (no https:// in WithEndpoint).
Go metrics cardinality:sdk/metric defaults to a cardinality limit of 2000; use WithCardinalityLimit(0) only when you intentionally need unlimited series, and expect otel.metric.overflow=true when the limit is hit.
.NET OTLP disk retry:OTEL_DOTNET_EXPERIMENTAL_OTLP_RETRY=disk now requires OTEL_DOTNET_EXPERIMENTAL_OTLP_DISK_RETRY_DIRECTORY_PATH; do not rely on an implicit temp-directory fallback.
