Observability Guide

What exists today

murmur-web wires Splunk RUM client-side (SplunkRumInit, fetching its own config from /api/v1/rum-config) for real-user monitoring, and murmur-web docs reference a Splunk O11y read path (src/lib/splunkO11yRead.ts).

UNVERIFIED

Server-side metrics/logs/traces tooling for murmur-api, murmur-intelligence, and murmur-audio wasn’t confirmed during this audit — this page documents what’s verified (client RUM) and flags the rest as open rather than describing dashboards that may not exist.

How service pages should show telemetry

Every SERVICE page carries a <Telemetry> block — a small metric readout row plus a dashboard link, not a full embedded dashboard (Atlas documents where the truth lives; it isn’t the observability tool itself):

—

P50 LATENCY

—

ERROR RATE

Atlas’s own observability

Per the build brief’s “don’t overbuild this” guidance: Atlas doesn’t stand up a parallel observability stack. Build failures surface in GitHub Actions (.github/workflows/ci.yml); once deployed to Cloudflare Pages/Workers, follow whatever Murmur’s existing Cloudflare monitoring pattern is for deploy health — none was found during this audit, so this is tracked as follow-up work in ATLAS_IMPLEMENTATION_REPORT.md rather than invented here.