Install Tracetest CLI (tracetest) and run tracetest configure to connect it to a running Tracetest server or Tracetest Cloud endpoint with YOUR_TOKEN
Create a test definition YAML specifying: trigger (HTTP request, grpc call, or Kafka message), a pollingProfile (how long to wait for the trace to be complete), and assertions written against span attributes
Write assertions using the selector syntax: select a span with a SpanSelector (e.g., span[tracing.system.name = "my-service"] and assert that http.response.status_code = 200 or db.statement contains the expected query
Run the test with tracetest run test -f test-definition.yaml and observe the structured output showing which span assertions passed or failed
Integrate the tracetest run command into a CI pipeline step after the application under test is deployed to a staging environment with a live trace backend (Jaeger, Tempo, or Datadog APM)
Use the test suite feature to chain multiple test definitions so that the output trace ID from one test can be referenced as context in the next
Known gotchas
Tracetest requires the trace to be fully ingested before assertions run; if your tracing backend has a high ingestion delay, increase the pollingProfile timeout or assertions will fail on incomplete traces
Assertions are evaluated against span attributes using the OpenTelemetry semantic conventions naming; ensure your instrumentation uses standard attribute names or write custom selectors against your actual attribute keys
Tracetest Cloud and self-hosted Tracetest have different CLI authentication flows; the configure command flags differ between versions — check the CLI version with tracetest version before running
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