Observability & debugging

Traces

Every run records a trace: each step that ran, in order, with its status, timing, input and output, and any error. Open it from the workflow panel (Runs → pick a run → Trace), and use the Inspector tab to look at one node's input and output in detail.

Run metrics

The Metrics tab summarises a run you choose: total duration, how many steps ran and their outcomes, the slowest nodes, and AI token use. For trends across many runs, use Analytics.

Compare runs

In the Compare tab, pick Run A and Run B (the two most recent runs are chosen by default) to see them side by side, including differences in duration and outcome.

Debugging checklist

• Find the failing step in the Trace and read its error.
• Check its input in the Inspector — is the data shaped the way the step expects?
• For AI steps, check Settings → AI Providers for a rejected or out-of-credit key.
• For external calls, check the integration or credential it uses, and the service's status page.
• Reproduce with a Dry Run to walk through the path without side effects.

Exporting

Workflows (and individual versions) can be exported as .fluxo.json files. Owners and admins can request a full workspace export (a zip with run history as CSV and JSON, the activity log, members and more) from Settings → Danger Zone; see Import & export. There are no scheduled exports or streaming to external log tools today.