If you run durable agents with tool approvals, approval policies now see the same tool definition fields as fresh calls. WorkflowAgent runs can also report as one traced operation via an OpenTelemetry ID generator bridge.
Key takeaways · AI-distilled
Tool approval continuations now restore server-defined tool metadata, dynamic status and title during validation, so approval policies receive the same tool definition fields as a new model call.
Durable `WorkflowAgent` runs can report as one telemetry operation covering the prompt, final output, steps and tools. Events recorded in the workflow are reported on completion or suspension with their original timestamps.
For OpenTelemetry, pass one `OpenTelemetryIdGenerator` instance as the tracer provider's idGenerator and sampler and to `new OpenTelemetry({ durable })` to carry trace context windowThe maximum amount of text a model can consider at once — its working memory for the current conversation or task.Full definition → across steps. Without that bridge, live per-model tracing is unchanged.
Terms in this piece · Glossary
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
