# Logs (/logs-debugging)





Open **Logs** to inspect a recorded workflow run. Start with its error, then follow the resolved inputs back to the block that supplied an unexpected value. The [Run logs reference](/logs-debugging/logging) explains filters, refresh, the editor console, and snapshots.

## Read a run trace [#read-a-run-trace]

Each run has a unique execution ID, a trigger, status, duration, and recorded cost in credits. Triggers include manual, API, schedule, webhook, MCP, and **Sim agent** for runs started through Chat.

Select a run to open **Log Details**. **Overview** contains run metadata and available workflow input, output, cost, and files. **Trace** shows timed spans for blocks, nested workflows, model requests, and tool calls. Access settings can hide trace, snapshot, or cost details.

<Image variant="detail" background="canyon" src="/static/docs-captures/log-trace.webp" alt="Trace for the summarizer workflow, showing Start and Summarize update with their durations" width="1036" height="650" />

Select a span to read its collapsible **Input**, **Output** or **Error**, and any **Thinking** or **Error message** sections. Input shows the values after references resolve; output shows what that step returned. Compare these values with the [data references](/workflows/data-flow) in your workflow.

Protected resolved secret values are masked in the log-facing copy. Workspace **Unredacted** exceptions and transformed values have different boundaries; see [Execution log protection](/platform/credentials#execution-log-protection).

A handled block error can appear in a successful run. For a failed run, find the unhandled error rather than assuming every error span caused the final status.

## Trace a failure backward [#trace-a-failure-backward]

1. Open the failed run and select its failed span in **Trace**. Read the error message.
2. Expand **Input** to see what the step received after references resolved.
3. Check required fields, types, and nested keys against the block's expected input.
4. Find the upstream block referenced by the incorrect input. Inspect its output and confirm that it ran.
5. Correct the reference, configuration, or upstream behavior, then run the workflow again.
6. Compare the new trace with the failed run. Confirm that the input is correct and the step completes successfully.

### Common failures [#common-failures]

| Failure | What to check |
| - | - |
| Missing input | A required field is empty or its reference resolves to no value. |
| Wrong reference | The block or output field named in the reference does not exist. |
| Type mismatch | A step receives a string where it expects an object, or another incompatible type. |
| Missing nested key | The object exists, but the requested field does not. |
| Empty agent message | An [Agent](/workflows/blocks/agent) returns no content for the next step to read. |
| External error | A tool or API returns an error; inspect its message and request inputs. |

Use a [Function](/workflows/blocks/function) block to reshape a value when the next step expects a different structure.

## Troubleshoot in Chat [#troubleshoot-in-chat]

When available on a failed run, select **Troubleshoot in Chat** to give Sim its debugging context. This action depends on Chat being enabled and the run supporting troubleshooting. Verify any proposed fix by rerunning the workflow and comparing traces.

## Retention [#retention]

Sim Cloud Free defaults to 30 days of log retention. Paid plans and billing-disabled deployments have no plan-based default expiry; configured policies can still remove older logs. See [Data retention](/platform/enterprise/data-retention) for the effective policy and cleanup behavior.

## Related docs [#related-docs]

* [Run logs](/logs-debugging/logging) for filters, details, console, and snapshots
* [Alerts](/logs-debugging/alerts) for failure and operational notifications
* [Data flow](/workflows/data-flow) for resolving block references