Observability
Stack Pack.

A multi-vendor pack — observability is one problem with many runtimes. OpenTelemetry as the protocol, Datadog and Grafana as the dominant vendors, Sentry for error monitoring, New Relic / Honeycomb / Splunk / Dynatrace as enterprise alternatives. Includes k6 for load testing and eBPF instrumentation patterns.

Why it's in the catalog. Most teams don't pick "an observability vendor" — they end up with three. The pack carries the integration patterns, the OTel semantic conventions, and the per-vendor opinion (Datadog for breadth, Grafana for control, Honeycomb for high-cardinality).

Headline products and features the pack carries opinions on. Each lives as its own canonical page under stacks/observability/<product>.md in the etyb-skills repo — currency-stamped, sourced, opinionated.

• OpenTelemetry (Collector + SDKs)
• OTel GenAI semantic conventions
• Datadog (APM + RUM + Logs + LLM Observability)
• Grafana Cloud (Loki + Mimir + Tempo + Pyroscope)
• Sentry
• New Relic
• Honeycomb
• Splunk
• Dynatrace
• k6
• Beyla (eBPF)

When /etyb routes a request to one of these roles AND the Observability pack is in play, the role's reference is composed with the platform-flavored knowledge from this stack. Same role, sharper answer.

• › sre-engineer
• › devops-engineer
• › system-architect
• › qa-engineer

The three things on Observability that a 2024-cutoff model is most likely to get wrong — and that this pack reflects.

• OTel GenAI semconv: cross-vendor LLM observability with consistent token + cost attributes
• Beyla eBPF: zero-instrumentation tracing for Go, Rust, Node, Python
• Grafana Pyroscope: continuous profiling integrated with the LGTM stack

The Observability pack comes with /etyb — one install, every pack ready to go. The overlay loads automatically when the work calls for it.

$ npx skills add e-t-y-b/etyb-skills

See the full install guide for Claude Code plugin and manual routes.