HyperProbe

Capture the value that the logs missed

HyperProbe addresses a specific debugging problem: a request took an unexpected branch, but existing logs do not show the value that caused it. A read-only probe can capture state when execution reaches a selected line without stopping the process.

This is most useful after narrowing the investigation to a service and code path. A probe will not return a useful capture if the affected request never reaches its location.

Get your first useful capture

1. Follow the quickstart to create a service and install the editor integration and appropriate SDK.
2. Initialize the SDK as directed and deploy the instrumented service with source and build metadata that match the running version.
3. Place a probe on an executable line near the suspected condition, then exercise that request path.
4. Inspect the captured variables and stack, or let your configured MCP client use that evidence to investigate the cause.

Separate setup from subsequent debugging

The no-redeploy workflow applies after instrumentation is in place. For a first trial, use a staging service and a reproducible request, then decide which captured fields are appropriate for your production environment. Free managed-cloud operation and enterprise private deployment have different data boundaries; choose the deployment that matches your requirements.