Add the spanmetrics connector to the Collector config, defining it as both a traces receiver and a metrics exporter
Configure the dimensions list to promote span attributes such as http.method and http.status_code into metric labels
Set histogram bucket boundaries appropriate for your service latency distribution
Wire the spanmetrics connector into a metrics pipeline that exports to Prometheus or your metrics backend
Use the generated request-rate, error-rate, and duration metrics to populate a golden-signals dashboard
Known gotchas
Each unique combination of dimension values creates a new metric series; adding high-cardinality span attributes as dimensions causes a metric cardinality explosion
The spanmetrics connector holds histogram state in memory; a Collector restart resets the histograms and produces a gap in cumulative metrics
Latency histograms use fixed bucket boundaries by default; if your service tail latency exceeds the highest configured bucket, all overflow is counted in the overflow bucket and you lose resolution
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp