Tracing
This page explains how AgentScope captures and visualizes the execution of MonadicSharp pipelines as distributed traces.
Work in Progress
This page is under construction. AgentScope is in active development.
Pipeline Steps as Spans
When this page is complete, it will explain how every step in a MonadicSharp pipeline — each Bind, Map, Match, and TryAsync call — is automatically instrumented as an OpenTelemetry span when MonadicSharp.Framework.Telemetry is configured. These spans carry the operation name, the success or failure outcome, any error details, and the elapsed time. AgentScope receives these spans via OTLP and groups them into a trace that represents the full lifecycle of a single pipeline execution.
Timeline Visualization
This section will describe the trace timeline view in the AgentScope dashboard. Each trace is rendered as a horizontal Gantt-style timeline where spans are positioned according to their start time and duration. Nested spans — representing sub-pipelines or delegated operations — are shown as child rows indented under their parent. Color coding distinguishes successful spans from failed ones at a glance, and hovering over a span reveals its full attribute payload.
Drill-Down on Errors
This section will cover the error drill-down workflow. When a pipeline step fails, AgentScope links the failure span to its full error context: the error type, the error message, the Result error value (if structured), and the stack trace if one was captured. From the timeline, a single click opens the error detail panel, allowing developers to understand exactly which step in a multi-stage AI pipeline failed and why, without needing to grep through logs.