See request and response bodies right in your traces

Feature Improvements Fixes

TL;DR

Trace summaries and exception views now show the full HTTP request and response bodies, so you can debug a failed call without leaving the trace. Plus 'did you mean?' suggestions in MCP and faster large-range tool calls.

See request and response bodies right in your traces

Debugging a failed request usually means jumping between the trace, your logs, and whatever captured the payload. Now the payload is in the trace itself.

Trace Details now has a Request & Response section in the Summary tab that renders the captured http.request.body and http.response.body — with syntax highlighting, copy, and show-more/less built in. The same bodies, along with the exception message, stacktrace, and HTTP context, show up in the expanded exception rows under Discover → Services. When a call fails, you can read exactly what was sent and what came back without leaving the trace.

Improvements

  • MCP
    • "Did you mean?" suggestions — mistype a service, host, or environment name and tools now return the closest matches instead of an empty result, so agents stop retrying bad input
    • Faster tool calls over large time ranges — fixed serial 5-minute chunks are replaced with parallel, adaptively-sized chunking, so queries return within client timeouts instead of hanging
  • Dashboards
    • Table panels can pivot multiple queries into separate columns, so you can build overview-style tables with entity names on the left and metrics, logs, and traces broken out across
    • Standardized empty, loading, and error states across detail pages, with consistent backgrounds and borders in both light and dark themes
  • Discover: Host processes, database monitoring, and LLM monitoring are now on for everyone — the feature flags are gone. The agent monitoring page now shows an integration link when it's empty, instead of a blank screen

Fixes

  • Dashboards
    • Charts now respect gaps in the data instead of drawing a continuous line or bar across missing points
    • Loading 7-day trend data no longer crashes the UI on service dashboards and other screens
  • Discover Hosts: CPU utilization in the host overview now matches the detail Metrics tab for the same host
  • RUM: The applications listing page no longer freezes the UI when opened
  • Alerting: Slack channel discovery now works in workspaces with a large number of channels, with a clearer "load more" affordance
  • User management: Inviting an existing user into a new workspace now correctly assigns their role