Automations and AI · how to

Read automation logs and receipts

Reconstruct automation runs from trigger context, published versions, step traces, receipts, and pending resumes.

3 min readUpdated August 12, 2026
For
Owner, Admin, Member
Availability
Your current plan and entitlements
Product evidence
WRKZY · reviewed August 12, 2026
Editorial review
WRKZY Editorial
Before you begin
  • Access to the automation and its run history
  • The approximate run time and a safe source-record identifier
  • The trigger, expected branch, and expected final state

Logs are the source of execution truth

The builder describes intent. Runs records observed execution. Open a workflow and select the Runs tab. Do not diagnose from an error toast alone; use the run, ordered step trace, visible action outcome, and any pending resume together.

Narrow the investigation

Use the time range, status, search, trigger, owner, step, and failure-reason filters. Start with the smallest window around the reported event. The page supports statuses including success, partial, and failed, and groups failure reasons so repeated operational issues can be separated from one-off record data.

Evidence in a run
EvidenceWhat it answersDo not assume
Trigger contextWhich event and saved input began the runThat the input was still current later
Published versionWhich saved workflow version ranThat the current builder draft matches it
Step traceWhich ordered path succeeded, failed, or skippedThat skipped always means an error
Action outcomeWhether a customer or external action completed, was safely reused, or remains uncertainThat a network timeout means nothing happened
Pending resumeWhich wait will continue, when, and from which stepThat changing the builder changes an existing wait

Investigate a reported failure

  1. Open Automations, choose the workflow, and select Runs.
  2. Set the time range and choose Failed or Partial.
  3. Search with a safe identifier or select the relevant owner, trigger, step, or failure reason.
  4. Expand the run that matches the approximate time.
  5. Read the run status, duration, trigger context, published version, and ordered trace.
  6. At the failed step, read its input summary, output summary, error, and receipt state.
  7. Check the waiting-work panel for pending, running, failed, or cancelled resumes tied to the run.
  8. Compare other failures in the same group before deciding whether the problem is configuration, data, connection health, or one record.

Interpret common patterns

  • Skipped with a safety reason: the control may be working as designed; inspect suppression, quiet hours, caps, or journey eligibility.
  • Partial: some steps completed and a later step did not. Never restart from the beginning without checking receipts.
  • Uncertain receipt: WRKZY blocks an unsafe automatic repeat because the external outcome cannot be proved.
  • Resume failed: the initial run reached a wait; the continuation failed later.
  • Several workflows failing on one channel: inspect Settings delivery or the connected provider before editing every workflow.

Run investigation path

  1. 01
    ScopeConfirm scope

    Identify the workspace, record, channel, time, and affected users.

  2. 02
    ObserveCheck visible state

    Read the current status instead of repeating the action.

  3. 03
    TraceReview history

    Use delivery events, audit history, or run logs to find the last good step.

  4. 04
    RecoverTry safe recovery

    Retry only when duplication and customer impact are understood.

  5. 05
    EscalateEscalate with redacted evidence

    Share expected versus actual results without secrets or excess customer data.

Confirm the scope, inspect visible state and history, try only a safe recovery, then escalate with redacted evidence.

Preserve and share evidence safely

Owners and Admins can use Export retained runs from the workflow menu when an operational review needs the retained dataset. Use Redact details when customer identity and step detail must be permanently removed; status, timing, version, and step outcomes remain for reporting. Redaction is irreversible.

For support, provide workspace slug, workflow name and ID, run ID, approximate time and timezone, published version, failed step label, receipt state, and redacted error. Never provide provider secrets, webhook keys, message bodies unrelated to the incident, passwords, or verification codes.

Expected result

You can state exactly what happened and whether a retry, individual wait recovery, provider fix, version rollback, or no action is appropriate.

Run diagnosis complete
  • Correct run and time window identified
  • Published version recorded
  • Failed or skipped path understood
  • Visible action outcome checked before retry
  • Pending continuation checked
  • Escalation evidence redacted
Was this guide useful?

Choose an answer. No message text or personal information is collected.

Still need help?

Contact WRKZY support with the workspace name and a safe, redacted example.

Contact support about this guide