​
Overview

How a run progresses, what updates in real time, and how to debug quickly when a step fails. Run overview

​
End‑to‑end lifecycle

  1. Start: click Run All (entire test) or Run Selected (subset). The header shows Running and disables editing.
  2. Session: the live browser connects or resets if needed.
  3. Execution: steps run in order; each step card moves from in‑progress to passed or failed.
  4. Artifacts: screenshots (per step), video (full session), trace (timeline), Raw Logs and AI Logs generate alongside execution.
  5. Completion: the run status is shown; open any step to review its artifacts.

​
What updates live

  • Step status: in‑progress, passed, failed; failing steps are highlighted.
  • Raw Logs: streaming technical output from the runner.
  • AI Logs: summarized, grouped insights for faster scanning.
  • Live preview (when available): reflects current page and interactions.

​
Debugging workflow (practical)

  1. Confirm the symptom:
    • Open the failing step’s screenshot: check visible state, key text, and URL.
    • If unclear, scrub the video around the failure time.
  2. Read what happened:
    • Open Raw Logs on the failing step: look for selector errors, timeouts, JS errors, or network issues.
    • Check AI Logs for a concise explanation and suggested next actions.
  3. Pinpoint timing or state:
    • Open Trace: review actions, waits, and network timeline around the failure.
    • Correlate timestamps between Trace and Raw Logs.
  4. Apply a fix:
    • Selector: use a more stable attribute or broader strategy.
    • Timing: replace fixed sleeps with explicit waits (visible, attached, navigation).
    • Data/flow: update environment variables, test data, or add a prerequisite snippet.
  5. Validate fast:
    • Use Run From Step to re‑execute from the failing point.
    • Do a full run when multiple assumptions changed.

​
Quick triage checklist

  • First failing step: fix earliest break; later errors often cascade.
  • Repeated failure on the same step: likely selector or timing fragility.
  • Rising run duration: check network calls and waits in Trace.
  • Confusing logs: read AI Logs first, then drill into Raw Logs.

​
Common scenarios and remedies

  • Element not found or detached: stabilize locator; wait for visible/attached.
  • Navigation timeout: ensure the action triggers navigation; wait for the correct event; verify target URL.
  • Dynamic content: prefer state‑based waits (text present, spinner gone) over fixed delays.
  • Data/env mismatch: confirm environment variables and inputs; log key values.