Page 5 of 8~112 min topic

Structured output lab

Debug accepting extra fields that change routing in the ticket triage schema gate

Page 5 reproduces and repairs the characteristic failure of the ticket triage JSON endpoint: accepting extra fields that change routing, or string priority 'high' vs enum.

~14 min this pageDebugging

1Learn the idea

Read

Reproduce before you repair

Do not start with a speculative fix for the ticket triage schema gate. Force the failure on purpose, save the before output, then change one cause at a time. Retries are allowed only for transient conditions—not for bad input that will fail forever on structured-output-lab.

Read

Force the failure

for (let attempt = 0; attempt <= maxRetries; attempt++) {
  try { return await operation(signal); }
  catch (error) {
    const failure = classify(error);
    if (!failure.retryable || attempt === maxRetries) throw failure;
    await sleep(jitteredBackoff(attempt), signal);
  }
}
throw new Error("unreachable");

Expected evidence: “Charged twice” becomes billing/high with a short summary; extra keys are rejected.. If you cannot reproduce on demand, you do not yet control the failure mode for structured-output-lab.

Read

Repair with a reviewable diff

After repair, rerun the exact reproduction command. Keep the failing fixture as a regression seed for the observability page. For the ticket triage JSON endpoint, remember the claim you are restoring: invalid model JSON is rejected; valid payloads pass a schema gate.

Read

Lab notebook: reproduce on command

Store a one-command reproduction for: accepting extra fields that change routing, or string priority 'high' vs enum. The command should use sample ticket + triage schema or a minimal mutant of it. Paste the failing output into notes/failure-before.txt (or your shell scrollback as copied text). After the fix, paste notes/failure-after.txt and keep both.

Retries belong only on transient faults. If the failure is bad input, a bad allowlist, or a logic bug in the ticket triage schema gate, retrying will amplify cost without repairing trust around return validated category, priority, summary, and confidence for a support ticket.

Read

Worked judgment

Classify the failure as prevent, detect, contain, or recover—using this lab’s language, not a generic poster. For structured-output-lab, the first fix should usually be detect+prevent at the boundary, because accepting extra fields that change routing, or string priority 'high' vs enum is cheaper to stop early than to explain in production prose.

Read

Why this stage matters for the ticket triage schema gate

At the debugging stage for structured-output-lab, the job is narrower than finishing a product demo. You are creating one progressive evidence piece about sample ticket + triage schema that later pages inherit without redefining success. Keep that fixture small enough to inspect by hand, keep outputs copy-pasteable as text, and refuse to narrate this baseline as if it were a production SLA: free-text model reply without schema validation.

For this page specifically, success looks like before/after evidence for the characteristic failure while still centering the user decision to return validated category, priority, summary, and confidence for a support ticket. If you cannot point to a file, command, or assertion that proves that for the ticket triage schema gate, stay on this page instead of advancing.

Glossary: tool · Glossary: structured output · Cheatsheet: production ops signals

Previous · Next

Go deeper

Before you start

Why this matters

Describe the smallest fixture that triggers accepting extra fields that change routing. Predict the first visible symptom (exception, wrong label, silent empty success). You will compare that prediction with the reproduction below.

In the wild

See how this idea shows up as a product and a company — then come back to the lesson. Skills transfer across vendors.

Check your understanding

Page assessment

Answer from memory. Completion is saved from this evidence, not from opening the next page.

1. Can you reproduce the failure with a one-command fixture?
2. Did you avoid retrying non-transient bad input?
3. Is before/after evidence saved as text (not only a screenshot)?
4. Does the repair restore the metric path toward: schema pass rate?

All responses are required.