> ## Documentation Index
> Fetch the complete documentation index at: https://evalgate.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Inspect a trace

> Start with what happened, then decompress spans, tools, cost, and related evidence.

# Inspect a trace

Open any row in **Evidence → Logs** (or a retained direct `/traces/{id}` link) to inspect one
production or staging interaction. The detail page leads with an
**investigation spine** (what happened → observed failure → related evidence), then the span/collection tabs.

![A screenshot of a failed checkout trace leading with Ended in error, Observed failure, and Related evidence.](https://mintcdn.com/evalgate/KV6BlM8ie_FdJpja/help/images/inspect-trace-investigation.png?fit=max\&auto=format\&n=KV6BlM8ie_FdJpja\&q=85\&s=55150049429baf93de8c8dc5e98b5351)

## What you see first

Before tabs, the page answers:

1. **Identity and outcome** — environment, release/version when known, and an
   honest execution badge (failed / completed / error). Completed without a
   measured evaluation is not labeled as Succeeded.
2. **Observed failure** — the concrete span or analysis signal that failed, when
   present. This is execution truth, not a behavioral pass/fail judgment.
3. **Related evidence** — session, failure pattern, evaluation run, candidate,
   review, or release links when those correlations exist. Missing links stay
   absent; EvalGate does not invent them.
4. **Prepared next actions** — for example **Label this failure** or **Create
   candidate** when the current evidence supports them.

Loop, seal, and lineage controls stay below the fold or in progressive disclosure.
Generic KPI card walls are not the lead.

## The tabs

| Tab | What it shows |
| - | - |
| **Tree** | The nested span hierarchy — the shape of the call |
| **Timeline** | Spans on a time axis, so you can see what ran in parallel and what blocked |
| **Thread** | The message turns between user, assistant, and tools |
| **Input / Output** | The exact request and response payloads |
| **Tools** | Every tool call, its arguments, result, and any error |
| **Scores** | Scorer outputs attached to this trace, with version and hash |
| **Cost** | Token usage and cost, linked to the model-call ledger |
| **Evidence** | Provenance state — complete or incomplete — and what is missing |
| **Review** | Any review task, decision, and rationale tied to this trace |

> [!NOTE]
>
> When a collection has more than 100 records, use its **Load more** control
> until the continuation is complete. The counts are exact, not estimates.

## What to look for next

1. **Observed failure / Tools** — confirm the failing span and whether a tool
   error masquerades as a model failure.
2. **Timeline** — find the long span. That is usually where time goes.
3. **Related evidence** — open a linked [Failure pattern](https://evalgate.com/docs/platform/failure-topics.md)
   or evaluation run when present.
4. **Cost** — a cost spike with few user-facing turns often means a hidden loop.
5. **Evidence** — if provenance is `incomplete`, this trace cannot be used as
   release evidence until the gap is fixed.

## Up next

[Save the view](https://evalgate.com/docs/help/traces/save-views.md) so you can return to this query. For named
recurring failures, continue in [Failure patterns](https://evalgate.com/docs/platform/failure-topics.md).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.