Workflow Runs
Inspect recent workflow runs, step outputs and errors.
Overview
Workflow Runs is the execution log for every automation you have built. It lists each run in reverse chronological order with its trigger source, start and finish time, total duration, final status, and a step-by-step breakdown of inputs, outputs, and errors. When something fails — a webhook returns 500, an AI call times out, a JSON path doesn't match — you can see exactly which step blew up and the data that flowed into it.
This is the tool you reach for when a workflow worked yesterday and isn't working today, or when you're tuning a new automation and want to confirm the right branches fired. It's most useful for developers, operations engineers, and anyone debugging webhook integrations, scheduled jobs, or multi-step automation pipelines.
How it works
Every workflow execution is persisted with a header row (workflow id, trigger type, queued/started/completed timestamps, status) and a child row per step that records the resolved input, the raw output, and any exception message and stack frame. Runs are produced by the Hangfire background job runner that powers workflow execution, so the log mirrors exactly what the worker saw at runtime — including retries.
The list view supports filtering by workflow, status (succeeded, failed, running, cancelled), and time range. Drilling into a single run expands the per-step panel so you can inspect the JSON envelope passed between steps and the error frame that stopped a failure.
Examples
- A scheduled "daily report" workflow shows Status: Failed at step 3 with
HttpRequestException: 401 Unauthorized— you rotated the API key but didn't update the secret. - A webhook-triggered run shows the inbound payload your partner system actually sent, including a stray trailing comma that broke the downstream JSON path.
- A manually triggered test run finishes in 1.4 s with the full output of each AI step visible for review before you wire it to production traffic.
- Filtering by Status: Running surfaces a stuck workflow that has been executing for two hours, letting you cancel and re-queue it.
FAQ
How long are run records kept?
Run history is retained per the platform's default retention policy. For long-term archival, export the run data or forward it to your own logging system from a workflow step.
Can I re-run a single failed step without re-running the whole workflow?
Not directly — you re-trigger the workflow from the start. Use a test trigger with the original input payload (visible in the run detail) to reproduce the failure quickly.
Do step outputs include secrets?
Secret values resolved through {{secrets.<name>}} are masked in the run log so credentials don't leak into history.
Why is a run stuck in Running?
Either the Hangfire worker is still processing a long step (e.g., a slow HTTP call), or the worker crashed mid-step. Check the step input timestamps to tell the two apart.