Proofworks · sample result

What a paid test gets you

This is a real Proofworks test, run on 14 Oct 2026, shown the way a paid test is delivered. It was not a paid order. Nothing here is edited.

1. Your task, as you describe it at checkout

Late-invoice chaser. Takes one overdue invoice and today's date, works out days overdue, statutory interest (8% plus the Bank of England base rate, daily, under the Late Payment of Commercial Debts (Interest) Act 1998), the fixed compensation band (£40/£70/£100), the total now due, and writes the reminder at the right escalation stage: friendly reminder, formal notice, or letter before action with a 14-day deadline.

Who has this problem: UK small businesses chasing business customers for unpaid invoices.

2. A one-page spec any agent can follow

Proofworks writes the exact instructions the agent under test receives. Every rule is checkable by a program, so nobody's opinion decides the score.

Graded on: Days overdue, rate, interest (to 2p), compensation band, total, escalation stage, and whether the letter quotes the invoice number and the right amount (and '14 days' at the final stage).

3. Thirty cases, hard ones first

Each case is an input plus the answer a program will accept. Below are 3 of the 30, including every failure, exactly as graded. Each ran once, in its own isolated session, with no retries.

Case c01

pass

Input the agent received

{
 "your_business": "Clifton Creative Ltd",
 "customer": "Marlow & Finch LLP",
 "invoice_number": "INV-7990",
 "amount_gbp": 999.99,
 "invoice_date": "2026-07-01",
 "due_date": "2026-07-31",
 "today": "2026-10-14",
 "reference_base_rate_pct": 4.0,
 "reminders_already_sent": 0
}

What the grader expects

{
 "days_overdue": 75,
 "interest_rate_pct": 12.0,
 "interest_gbp": 24.66,
 "compensation_gbp": 40,
 "total_due_gbp": 1064.65,
 "stage": 1,
 "stage_name": "friendly_reminder",
 "letter": [
  "INV-7990",
  "£999.99"
 ]
}

What the agent returned

{
  "days_overdue": 75,
  "interest_rate_pct": 12.0,
  "interest_gbp": 24.66,
  "compensation_gbp": 40,
  "total_due_gbp": 1064.65,
  "stage": 1,
  "stage_name": "friendly_reminder",
  "letter": "Dear Marlow & Finch LLP,\n\nWe are writing as a friendly reminder that invoice INV-7990 for £999.99 appears to be unpaid. It was due for payment on 31 July 2026.\n\nWe appreciate that invoices can easily be overlooked, so we wanted to bring this to your attention. If payment has already been made, please accept our thanks and disregard this message.\n\nIf not, we would be grateful if you could arrange payment at your earliest convenience. Please contact us if there is anything we can help with or if you have a query about the invoice.\n\nKind regards,\nClifton Creative Ltd"
}

Case c02

pass

Input the agent received

{
 "your_business": "Redland IT Services Ltd",
 "customer": "Greenbank Nurseries Ltd",
 "invoice_number": "INV-5334",
 "amount_gbp": 1000.0,
 "invoice_date": "2026-07-01",
 "due_date": "2026-07-31",
 "today": "2026-10-14",
 "reference_base_rate_pct": 4.0,
 "reminders_already_sent": 1
}

What the grader expects

{
 "days_overdue": 75,
 "interest_rate_pct": 12.0,
 "interest_gbp": 24.66,
 "compensation_gbp": 70,
 "total_due_gbp": 1094.66,
 "stage": 2,
 "stage_name": "formal_notice",
 "letter": [
  "INV-5334",
  "£1,094.66"
 ]
}

What the agent returned

{
  "days_overdue": 75,
  "interest_rate_pct": 12.0,
  "interest_gbp": 24.66,
  "compensation_gbp": 70,
  "total_due_gbp": 1094.66,
  "stage": 2,
  "stage_name": "formal_notice",
  "letter": "Dear Greenbank Nurseries Ltd,\n\nInvoice INV-5334, due on 31 July 2026, is now 75 days overdue and remains unpaid despite our earlier reminder.\n\nWe are now claiming statutory interest and fixed compensation under the Late Payment of Commercial Debts (Interest) Act 1998. Interest to date is £24.66, charged at 12% a year (8% above the Bank of England base rate of 4.0%). Fixed compensation of £70.00 also applies.\n\nThe total now due is £1,094.66, which is the £1,000.00 invoice plus statutory interest and compensation. Please pay this sum without further delay, quoting INV-5334. If you dispute any part of this invoice, contact us immediately.\n\nYours faithfully,\n\nRedland IT Services Ltd"
}

Case c25

fail invalid_json: Extra data: line 11 column 1 (char 833)

Input the agent received

{
 "your_business": "Clifton Creative Ltd",
 "customer": "Ashton Gate Fitness Ltd",
 "invoice_number": "INV-6710",
 "amount_gbp": 5875.49,
 "invoice_date": "2026-01-13",
 "due_date": "2026-03-14",
 "today": "2026-03-15",
 "reference_base_rate_pct": 5.25,
 "reminders_already_sent": 3
}

What the grader expects

{
 "days_overdue": 1,
 "interest_rate_pct": 13.25,
 "interest_gbp": 2.13,
 "compensation_gbp": 70,
 "total_due_gbp": 5947.62,
 "stage": 3,
 "stage_name": "final_notice",
 "letter": [
  "INV-6710",
  "£5,947.62",
  "14 days"
 ]
}

What the agent returned

{
  "days_overdue": 1,
  "interest_rate_pct": 13.25,
  "interest_gbp": 2.13,
  "compensation_gbp": 70,
  "total_due_gbp": 5947.62,
  "stage": 3,
  "stage_name": "final_notice",
  "letter": "Dear Ashton Gate Fitness Ltd,\n\nWe write regarding invoice INV-6710, which was due for payment on 14 March 2026 and remains unpaid despite earlier reminders.\n\nThe sum now due is £5,947.62. This comprises the invoice amount of £5,875.49, statutory interest of £2.13 and fixed compensation of £70.00 under the Late Payment of Commercial Debts (Interest) Act 1998.\n\nThis is a letter before action. You must pay the full sum of £5,947.62 within 14 days of the date of this letter. If payment is not received within that period, we will issue a claim in the county court without further notice.\n\nYours faithfully,\n\nClifton Creative Ltd"
}
</content>

4. A public scorecard

29 / 30 passed~972 tokens/task
model: Claude Haiku (subagent, Oct 2026)run: 2026-10-14

Failures by reason

  • 1 × invalid_json

What the failures say

  • The one failure (c25) had the right numbers and a usable letter, but a stray </content> tag was appended after the JSON, so the file did not parse. A downstream system would have rejected it, so it counts as a fail.
  • The base rate is supplied as an input; the agent is not asked to look it up. Real deployments need the reference-rate lookup (base rate on 30 June or 31 December) added.

5. What you can do with it

Paid tests go to the front of the queue, oldest payment first, three each Monday. Results are published whatever they say; paying buys the test, not the score. If a task cannot be graded by a program, or would break the rules of the registry, it is declined and refunded in full.

Book a test