Which signal is most useful for debugging *why* an agent picked a wrong tool?

The state snapshot and routing decision at the supervisor node.

The size of the prompt template.

Tracing agent runs — Semantic Web Academy

Overview

Spans per node, token cost per step, latency per edge — LangSmith-style.

Why it matters

An agent is a distributed system. Structured tracing is the only way to debug the graph of calls, not just one LLM response at a time.

Where this sits in the stack

Tracing agent runs is one of the load-bearing decisions in a KG/RAG/agent system: choices made here propagate to retrieval quality, agent reliability, cost per query, and the on-call burden of whoever ships it. Teams that name this trade-off explicitly ship faster than teams that leave it implicit.

Why this is load-bearing

Tracing agent runs is the building-code of this layer. You can ignore building codes on a shed, but the moment you put two storeys on top of the same foundation they decide whether the structure stands or falls. In a KG/RAG/agent stack, the equivalent of 'two storeys' is the second feature you ship on top of this primitive — GraphRAG on top of chunking, supervisor agents on top of state machines, regression CI on top of metrics. The cost of cutting the wrong corner now is paid by every later layer, with interest.

An agent is a distributed system. Structured tracing is the only way to debug the graph of calls, not just one LLM response at a time.

Reflect — apply it

Anchor tracing agent runs to something concrete in your own work.

›Where have you seen tracing agent runs done well? Name one team or product and what they got right.
›Where have you seen it done badly? What was the first symptom that surfaced (latency, hallucination, cost, outage)?
›What is the *cheapest* version of this you could ship in your next sprint, and what single metric would tell you it's working?

Reading in progress · 0 of 3 activities done