Runs & results
A run is any execution Cofactor performs against your application. Every run captures the same kind of evidence — statuses, steps, assertions, screenshots, video, and timing — and feeds the same results layer of issues and triage. This page explains the run types, how to choose between them, and where their results go.
01Run types
| Type | What it is | Start it when |
|---|---|---|
| Workflow run | A single execution of one workflow. The atomic run record — everything else builds on it. | You want the result of one specific flow. |
| Test run | A grouped execution of many workflows at once (a test suite, a tag, or all published flows). | You want regression coverage across your suite. |
| Ad-hoc test | An agent-driven, one-off test from custom instructions — no saved workflow. | You want to explore or validate something without authoring a flow first. |
A test run is made of workflow runs: open a test run, drill into a row, and you land on the underlying workflow run with its full detail view. An ad-hoc test produces its own run record with the same evidence, but isn't tied to a reusable workflow.
02The run lifecycle
Every run moves through the same states, whatever its type:
| Status | Meaning |
|---|---|
| Queued | Waiting to start. |
| Running | Execution in progress. |
| Waiting | Paused at a Wait node; resumes automatically. |
| Success | Completed with no failing steps or assertions. |
| Failed | Ended with one or more execution-breaking failures. |
| Canceled | Stopped manually. |
Within a run, each step (one workflow node) records its own status, timing, and assertion results. Assertions resolve to Pass, Warn (needs review, doesn't fail the run), or Fail (marks the step/run failed). See Workflow runs for how to read this evidence and investigate failures.
03Where results go
Runs don't just pass or fail — they produce a results layer you act on:
- Issues — when a run surfaces a real defect, Cofactor files a deduplicated issue with the screenshot, steps, and evidence.
- Triage & diagnostics — when a test run has multiple failures, Cofactor clusters them by likely root cause and attaches per-failure diagnostics so you investigate causes, not symptoms.