Pre-launch

Creditcrest Decisioning

A lender’s own credit policy, versioned and replayable against the version that was in force on the day.

An example. Every figure in it is invented.

Every figure in this example is invented. There is no borrower, no bank statement and no lender: the transactions below were generated from the numbers in the form so that the policy engine has a statement to read, which is what it reads.

The credit policy replayed here is the one that says this at the top of itself. models/policies/example-au.txt is an example. Its thresholds are round numbers chosen to make the mechanism visible. They are not advice, not a market standard and not anybody’s policy. A real policy is written by a lender's credit risk function against their own book and signed off by people whose names go on it.

Replaying round numbers — models/policies/example-au.txt. Written to make the mechanism visible: a $300 surplus floor, a 45 per cent debt service ceiling, thresholds a reader can do in their head while watching the replay work. Or run the same application against In the shape of a published guideline.

Dollars a month. The monthly pay credited to the account.

Dollars a month. Monthly income of a kind that is not wages — rental, dividends, side work — which a lender counts at less than its face value.

Dollars a month. The monthly rent.

Dollars a month. Monthly card spending on groceries, fuel and utilities.

Dollars a month. Monthly spending with a wagering operator.

Dollars a month. The lender’s own monthly expense benchmark for this household. Or “none”, which is what a lender who has not supplied a benchmark should type: the assessment is then incomplete and the whole policy run is unresolved, which is the engine refusing to report conditions met on evidence that may understate expenses.

Dollars a month. What the applicant said they spend each month.

Dollars. Every debt the lender counts, including the loan applied for, existing loans and the full limit of every card.

Dollars a year. Gross annual income before tax, which is what a debt-to-income ratio is measured against and is not what a bank statement shows.

Dollars. The amount applied for.

Per cent a year. The annual rate offered, before the assessment buffer.

Years. The term applied for.

YYYY-MM-DD. The day the application was assessed, which decides which version of the policy it is replayed against.

Neither of them is anybody's policy, and both say so. Changing this also changes the day the example defaults to being assessed on, because the two policies changed on different days.

The expense benchmark is the lender's own. This repository holds no benchmark table and will not ship one: the Household Expenditure Measure is licensed from the Melbourne Institute, and inventing plausible benchmark figures would be the most damaging thing this software could do, because they would look right.

What the replay found

3 conditions came to a different answer on 2026-09-26 than on 2026-05-31, on the same evidence, because the clause behind them changed between 2026.03 and 2026.06.

Both runs were handed the same reading, the same serviceability assessment and the same proposed loan. Those three together hash to 26dcf681de92fa4804682248cc91eb695b8cab789306ed057e7da98210316aaa, computed by the same canonical() and SHA-256 that seal a record, so “nothing but the policy changed” is something you can check rather than something this page asserts.

Debt to income, and whose threshold it is

The cap APRA sets is on the lender's book. The threshold in this policy is the lender's own.

From February 2026 APRA limits an authorised deposit-taking institution to writing no more than 20 per cent of its new residential mortgage lending, measured each quarter and counted separately for owner-occupier and investor lending, at a debt-to-income ratio of six or more. That caps the mix of a lender’s own book. It does not prohibit a borrower: it does not say that a person at a ratio of six may not be lent to, and no threshold on this surface is APRA’s rule. What a lender does about a portfolio limit — a ceiling, a referral band, a delegation level, or nothing at all — is that lender’s own credit policy, which is the thing versioned and hashed here.

Two lenders with the same applicant in front of them may write that policy differently, and a lender close to its quarterly cap may write it differently in September than it did in June. That is the thing this surface holds, and it is the reason a threshold needs a date on it.

The example policy replayed here is invented consumer lending, which APRA's mortgage measure does not reach at all. models/policies/example-au.txt is an example. Its thresholds are round numbers chosen to make the mechanism visible. They are not advice, not a market standard and not anybody’s policy. Debt to income is in the example because it is a threshold lenders are moving this year, and a threshold that moves is exactly what a replay is for.

This ratio was supplied. This software does not compute one, and the reason is not a limitation. A debt-to-income ratio is measured against gross income, and a bank statement shows what was credited to an account, which is take-home pay; there is no tax table in this repository and there will not be one. And what counts as debt — a card at its limit or at its balance, a HELP debt, a guarantee — is a definition the lender writes. Both of those belong to the lender, so the ratio arrives as an input with the two figures behind it and the engine reports what it was handed.

Debt counted
$495,000.00 — including the $18,000.00 applied for
Gross annual income
$75,000.00 before tax, which is not the $4,800.00 a month credited to the account
Debt to income
6.6 times — $495,000.00 ÷ $75,000.00
Where it came from
the two figures typed into the form above

“Debt to income” and “loan to income” are two different figures and this policy carries both. Loan to income is the principal applied for as a multiple of assessed annual income, and the engine works it out from the application and the assessment. Debt to income is every debt the lender counts, this loan included, over gross annual income. A policy written against the wrong one of those would look entirely reasonable on the page.

A referral band reads backwards the first time. A refer line states the condition under which the application does not go to a person. So a lender who wants a ratio in the band this policy refers at looked at by a human writes it as the requirement to be outside that band:

debt to income not between 5.5 and 6.5

Both ends of a band are included, which is what “between” means when a person says it out loud. The format is documented in the policy file itself, where the person editing it will be.

The same application, under both versions

As assessed on 31 May 2026

2026.03, in force from 1 March 2026 up to and including 31 May 2026, having been retired on 1 June 2026.

Rules hashed to 8400e4ccd45d6d4696e20f87c41cc6996ab7b9dc25bd429cc5c32316c4727caa

All 9 conditions in 2026.03 were met on the evidence supplied. 2 tests send this application to a person to look at.

As the policy stands, 26 September 2026

2026.06, in force from 1 June 2026 and still in force.

Rules hashed to 083754ea2ff07e44c7b5d23295917b5fdadea8bee580c149de1d2a1cb155bb90

2 conditions were not met. 1 test sends this application to a person to look at.

As assessed on 31 May 2026

Conditions met
12
Not met
0
To a person
2
Could not be read
0

As the policy stands, 26 September 2026

Conditions met
11
Not met
2
To a person
1
Could not be read
0

A referral is not a failed condition. A refer test that bites sends the application to a person to look at; it is counted in its own box and marked on its own row, because counting it as a failure would report a policy as breached when the policy said "somebody should see this". The four boxes count every test in the version, so "conditions met" includes the referral tests that did not bite; the engine's own sentence above counts only the conditions that must be met, which is why the two numbers differ.

The 3 clauses that changed the answer

surplus — the policy said monthly surplus at least $300 and now says monthly surplus at least $600.

On 31 May 2026 that condition was Met; on 26 September 2026, on the same evidence, it is Not met. Not met. monthly surplus is $463.22. This rule requires at least $600. A surplus at the assessed rate is what the applicant has left if the rate moves against them. Raised in June after a portfolio review.

dti-ceiling — the policy said debt to income at most 7 and now says debt to income at most 6.

On 31 May 2026 that condition was Met; on 26 September 2026, on the same evidence, it is Not met. Not met. debt to income is 6.6 times. This rule requires no more than 6 times. Every debt this lender counts, including the loan applied for, as a multiple of gross annual income. Lowered from seven in June. This is the lender's own ceiling. It is not APRA's cap, which limits the share of a lender's new mortgage book written above a ratio and says nothing about whether any one borrower may be lent to.

dti-band — the policy said debt to income not between 6.5 and 7 and now says debt to income not between 5.5 and 6.5.

On 31 May 2026 that condition was Not met; on 26 September 2026, on the same evidence, it is Met. Met. debt to income is 6.6 times. This referral test requires it to be outside the band 5.5 times to 6.5 times.

One clause that changed and did not change an answer

Worth printing rather than dropping: a lender who moved a threshold and saw nothing happen on this application has still moved it, and it will matter on the next one.

loan-to-income — “loan to income at most 6” became “loan to income at most 5”. Met. loan to income is 0.31 times. This rule requires no more than 5 times.

What the policy's author wrote about the change

June: the surplus floor moved from $300 to $600, the loan-to-income ceiling from 6 to 5, the debt-to-income ceiling from 7 to 6 and the referral band under it from 6.5-7.0 to 5.5-6.5, after a portfolio review. An application assessed in May is still judged against 2026.03, which is the whole reason the versions carry dates.

That is the author's account of it. The clauses above are the fact, read from the rules themselves, so a change nobody wrote a note about still appears.

Between 2026.03 and 2026.06: 4 changed, 9 unchanged.

What the evidence says about itself

The statement shows about $1,758.47 a month of living expenses against $1,400 declared. The larger figure was used, and the difference is worth asking about. (from the serviceability assessment)

Every condition, both versions, with what it read

Every rule that ran, in the order the policy wrote them: the clause it came from in the policy's own words, the figure it read, the value it compared that figure against, and the transactions behind the figure. A decision a lender cannot read back to a regulator line by line is the thing this product exists to prevent, so nothing here is summarised away.

evidence-complete this engine’s own condition, not the policy’s

What it measures: Whether every input a complete serviceability assessment needs was actually supplied

As assessed on 31 May 2026

Met assessment complete is true

Met. The serviceability assessment reports that every input a complete assessment needs was supplied.

As the policy stands, 26 September 2026

Met assessment complete is true

Met. The serviceability assessment reports that every input a complete assessment needs was supplied.

Why the policy has this condition, in the policy's own words: Rules below read figures from the serviceability assessment. If that assessment is itself incomplete, so is every conclusion drawn from it.

What this condition read, and what it compared it against

The figure this condition read was true.

The figures behind it

Gaps
1
Complete
yes

This figure comes out of the serviceability arithmetic. Every step of it, with its inputs and the rule that produced it, is under “The arithmetic” below.

surplus

What it measures: What is left each month once assessed income has met living expenses, commitments and the proposed repayment

As assessed on 31 May 2026

Met monthly surplus at least $300

Met. monthly surplus is $463.22. This rule requires at least $300.

As the policy stands, 26 September 2026

Not met monthly surplus at least $600

Not met. monthly surplus is $463.22. This rule requires at least $600. A surplus at the assessed rate is what the applicant has left if the rate moves against them. Raised in June after a portfolio review.

The clause changed. It read “monthly surplus at least $300” and now reads “monthly surplus at least $600”. That change is what moved the answer.

Why the policy had this condition, as assessed on 31 may 2026: A surplus at the assessed rate is what the applicant has left if the rate moves against them. A figure near zero is arithmetically true and nobody should lend on it.

Why it has it as the policy stands, 26 september 2026: A surplus at the assessed rate is what the applicant has left if the rate moves against them. Raised in June after a portfolio review.

What this condition read, and what it compared it against

The figure this condition read was $463.22.

The figures behind it

Assessed income
4,799.99
Living expenses
2,700
Commitments
1,200
Proposed repayment
436.77

This figure comes out of the serviceability arithmetic. Every step of it, with its inputs and the rule that produced it, is under “The arithmetic” below.

surplus-ratio

What it measures: The monthly surplus as a share of assessed income

As assessed on 31 May 2026

Met net surplus ratio at least 5%

Met. net surplus ratio is 9.65%. This rule requires at least 5%.

As the policy stands, 26 September 2026

Met net surplus ratio at least 5%

Met. net surplus ratio is 9.65%. This rule requires at least 5%.

Why the policy has this condition, in the policy's own words: The same question as a share of income, so that a large income with large outgoings is not read as comfortable.

What this condition read, and what it compared it against

The figure this condition read was 9.65%.

The figures behind it

Assessed income
4,799.99
Living expenses
2,700
Commitments
1,200
Proposed repayment
436.77

This figure comes out of the serviceability arithmetic. Every step of it, with its inputs and the rule that produced it, is under “The arithmetic” below.

debt-service

What it measures: All repayments, existing and proposed, as a share of assessed income

As assessed on 31 May 2026

Met debt service ratio at most 45%

Met. debt service ratio is 34.1%. This rule requires no more than 45%.

As the policy stands, 26 September 2026

Met debt service ratio at most 45%

Met. debt service ratio is 34.1%. This rule requires no more than 45%.

Why the policy has this condition, in the policy's own words: Every repayment, existing and proposed, measured against assessed income.

What this condition read, and what it compared it against

The figure this condition read was 34.1%.

The figures behind it

Assessed income
4,799.99
Living expenses
2,700
Commitments
1,200
Proposed repayment
436.77

This figure comes out of the serviceability arithmetic. Every step of it, with its inputs and the rule that produced it, is under “The arithmetic” below.

loan-to-income

What it measures: The principal as a multiple of assessed annual income

As assessed on 31 May 2026

Met loan to income at most 6

Met. loan to income is 0.31 times. This rule requires no more than 6 times.

As the policy stands, 26 September 2026

Met loan to income at most 5

Met. loan to income is 0.31 times. This rule requires no more than 5 times.

The clause changed. It read “loan to income at most 6” and now reads “loan to income at most 5”. The answer did not move: on this application the figure sits on the same side of both thresholds.

Why the policy had this condition, as assessed on 31 may 2026: The principal as a multiple of assessed annual income.

Why it has it as the policy stands, 26 september 2026: The principal as a multiple of assessed annual income. Lowered in June.

What this condition read, and what it compared it against

The figure this condition read was 0.31 times.

The figures behind it

Principal
18,000
Assessed annual income
57,599.88

This figure comes out of the serviceability arithmetic. Every step of it, with its inputs and the rule that produced it, is under “The arithmetic” below.

dti-ceiling

What it measures: Every debt the lender counts, including the loan applied for, as a multiple of gross annual income — supplied, because gross income is not on a bank statement and what counts as debt is the lender’s own definition

As assessed on 31 May 2026

Met debt to income at most 7

Met. debt to income is 6.6 times. This rule requires no more than 7 times.

As the policy stands, 26 September 2026

Not met debt to income at most 6

Not met. debt to income is 6.6 times. This rule requires no more than 6 times. Every debt this lender counts, including the loan applied for, as a multiple of gross annual income. Lowered from seven in June. This is the lender's own ceiling. It is not APRA's cap, which limits the share of a lender's new mortgage book written above a ratio and says nothing about whether any one borrower may be lent to.

The clause changed. It read “debt to income at most 7” and now reads “debt to income at most 6”. That change is what moved the answer.

Why the policy had this condition, as assessed on 31 may 2026: Every debt this lender counts, including the loan applied for, as a multiple of gross annual income. This is the lender's own ceiling. It is not APRA's cap, which limits the share of a lender's new mortgage book written above a ratio and says nothing about whether any one borrower may be lent to.

Why it has it as the policy stands, 26 september 2026: Every debt this lender counts, including the loan applied for, as a multiple of gross annual income. Lowered from seven in June. This is the lender's own ceiling. It is not APRA's cap, which limits the share of a lender's new mortgage book written above a ratio and says nothing about whether any one borrower may be lent to.

What this condition read, and what it compared it against

The figure this condition read was 6.6 times.

The figures behind it

Total debt
495,000
Gross annual income
75,000
Counted by
the figures typed into the form on this page, which is the lender’s own definition of what counts

history

What it measures: How many days the statement covers, first transaction to last

As assessed on 31 May 2026

Met statement days at least 90

Met. statement days is 176. This rule requires at least 90.

As the policy stands, 26 September 2026

Met statement days at least 90

Met. statement days is 176. This rule requires at least 90.

Why the policy has this condition, in the policy's own words: Three months is the shortest period in which a fortnightly rhythm can be seen at all.

What this condition read, and what it compared it against

The figure this condition read was 176.

The figures behind it

From
2025-12-05
To
2026-05-29

coverage

What it measures: The share of transactions this software could identify

As assessed on 31 May 2026

Met categorised share at least 80%

Met. categorised share is 100%. This rule requires at least 80%.

As the policy stands, 26 September 2026

Met categorised share at least 80%

Met. categorised share is 100%. This rule requires at least 80%.

Why the policy has this condition, in the policy's own words: Below this, too much of the statement is unread for the expense picture to mean anything. The reading lists the payees it could not identify.

What this condition read, and what it compared it against

The figure this condition read was 100%.

The figures behind it

Categorised
77
Transactions
77

income

What it measures: How many separate regular income streams were identified

As assessed on 31 May 2026

Met income streams at least 1

Met. income streams is 1. This rule requires at least 1.

As the policy stands, 26 September 2026

Met income streams at least 1

Met. income streams is 1. This rule requires at least 1.

Why the policy has this condition, in the policy's own words: An assessment with no identified regular income is an assessment of nothing.

What this condition read, and what it compared it against

The figure this condition read was 1.

The figures behind it

Verified monthly
4,799.99

The 1 row behind the figure

Where a row is a recurring arrangement rather than a single transaction, the date is its first payment and the amount is the monthly figure it was read as.

DateDescriptionAmount
10 December 2025 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $4,799.99

dti-band sends it to a person

What it measures: Every debt the lender counts, including the loan applied for, as a multiple of gross annual income — supplied, because gross income is not on a bank statement and what counts as debt is the lender’s own definition

As assessed on 31 May 2026

To a person debt to income not between 6.5 and 7

Not met. debt to income is 6.6 times. This referral test requires it to be outside the band 6.5 times to 7 times. A ratio in the band below this lender's ceiling goes to a person rather than through. Read the polarity carefully: a refer states the condition under which the application does NOT go to a person, so the band a person looks at is the one written as "not between".

As the policy stands, 26 September 2026

Met debt to income not between 5.5 and 6.5

Met. debt to income is 6.6 times. This referral test requires it to be outside the band 5.5 times to 6.5 times.

The clause changed. It read “debt to income not between 6.5 and 7” and now reads “debt to income not between 5.5 and 6.5”. That change is what moved the answer.

Why the policy had this condition, as assessed on 31 may 2026: A ratio in the band below this lender's ceiling goes to a person rather than through. Read the polarity carefully: a refer states the condition under which the application does NOT go to a person, so the band a person looks at is the one written as "not between".

Why it has it as the policy stands, 26 september 2026: A ratio in the band below this lender's ceiling goes to a person rather than through. The band was 6.5 to 7.0 in March and was lowered with the ceiling in June. Read the polarity carefully: a refer states the condition under which the application does NOT go to a person, so the band a person looks at is the one written as "not between".

What this condition read, and what it compared it against

The figure this condition read was 6.6 times.

The figures behind it

Total debt
495,000
Gross annual income
75,000
Counted by
the figures typed into the form on this page, which is the lender’s own definition of what counts

inferred-income sends it to a person

What it measures: The share of verified income that comes from credits whose description does not say they are income

As assessed on 31 May 2026

Met inferred income share at most 50%

Met. inferred income share is 0%. This referral test requires no more than 50%.

As the policy stands, 26 September 2026

Met inferred income share at most 50%

Met. inferred income share is 0%. This referral test requires no more than 50%.

Why the policy has this condition, in the policy's own words: Income counted because it repeats, from credits whose description does not name a payer, should be confirmed with the applicant before it carries an application.

What this condition read, and what it compared it against

The figure this condition read was 0%.

The figures behind it

Inferred monthly
0
Verified monthly
4,799.99

gambling sends it to a person

What it measures: Monthly spending in the gambling category

As assessed on 31 May 2026

To a person spend on gambling at most $100

Not met. spend on gambling is $155.71. This referral test requires no more than $100. Not a reason in itself. It is a fact a person should see beside the rest of the file.

As the policy stands, 26 September 2026

To a person spend on gambling at most $100

Not met. spend on gambling is $155.71. This referral test requires no more than $100. Not a reason in itself. It is a fact a person should see beside the rest of the file.

Why the policy has this condition, in the policy's own words: Not a reason in itself. It is a fact a person should see beside the rest of the file.

What this condition read, and what it compared it against

The figure this condition read was $155.71.

The figures behind it

Total
899.99
Count
13
Months
5.78

The 13 rows behind the figure

Where a row is a recurring arrangement rather than a single transaction, the date is its first payment and the amount is the monthly figure it was read as.

DateDescriptionAmount
11 December 2025 SPORTSBET DEPOSIT -$69.23
25 December 2025 SPORTSBET DEPOSIT -$69.23
8 January 2026 SPORTSBET DEPOSIT -$69.23
22 January 2026 SPORTSBET DEPOSIT -$69.23
5 February 2026 SPORTSBET DEPOSIT -$69.23
19 February 2026 SPORTSBET DEPOSIT -$69.23
5 March 2026 SPORTSBET DEPOSIT -$69.23
19 March 2026 SPORTSBET DEPOSIT -$69.23
2 April 2026 SPORTSBET DEPOSIT -$69.23
16 April 2026 SPORTSBET DEPOSIT -$69.23
30 April 2026 SPORTSBET DEPOSIT -$69.23
14 May 2026 SPORTSBET DEPOSIT -$69.23
28 May 2026 SPORTSBET DEPOSIT -$69.23

small-credit sends it to a person

What it measures: Whether the reading raised the small_amount_credit signal

As assessed on 31 May 2026

Met signal small_amount_credit is absent

Met. signal small_amount_credit is absent. This referral test requires the signal to be absent.

As the policy stands, 26 September 2026

Met signal small_amount_credit is absent

Met. signal small_amount_credit is absent. This referral test requires the signal to be absent.

Why the policy has this condition, in the policy's own words: Payments to businesses whose products are commonly small amount credit contracts. The rebuttable presumptions of unsuitability that used to attach to this were repealed in 2022, so this referral is this lender's own judgement and not a requirement of the Act.

What this condition read, and what it compared it against

The figure this condition read was absent.

dishonours sends it to a person

What it measures: Whether the reading raised the dishonours signal

As assessed on 31 May 2026

Met signal dishonours is absent

Met. signal dishonours is absent. This referral test requires the signal to be absent.

As the policy stands, 26 September 2026

Met signal dishonours is absent

Met. signal dishonours is absent. This referral test requires the signal to be absent.

Why the policy has this condition, in the policy's own words: Dishonour and overdrawn fees say something about the month to month that a monthly average does not.

What this condition read, and what it compared it against

The figure this condition read was absent.

The arithmetic

Every step of the serviceability assessment both runs read, with its inputs and the rule that produced it. The surplus, the two ratios and the loan-to-income multiple in the conditions above all come out of this ledger and out of nothing else. The debt-to-income ratio does not, and it is listed separately below for that reason: it was supplied, because gross income is not on a bank statement and what counts as debt is the lender's own definition.

StepWhatAmountThe rule that produced it
income DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $4,799.99 Counted in full
expenses Living expenses -$2,700.00 The highest of the available figures, which here was the lender’s benchmark for this household
commitment RENT HARCOURTS ELSTERNWICK REF 88210 -$1,200.00 Found on the statement, 6 payments monthly
proposed The proposed personal loan -$436.77 Assessed at 15.90% — the offered rate plus a 3.0% buffer, with a 5.25% floor
Assessment rate
15.90% — the 12.9% offered plus the buffer, with a floor
Assessed income
$4,799.99
Living expenses used
$2,700.00 — the highest of declared $1,400.00, benchmark $2,700.00 and what the statement showed
Commitments
$1,200.00
Repayment on the proposed loan
$436.77
Monthly surplus
$463.22
Net surplus ratio
9.65%
Debt service ratio
34.10%
Every input a complete assessment needs
supplied
Debt to income — supplied, not from the ledger
6.6 times — $495,000.00 of debt over $75,000.00 of gross annual income
What the assessment says is thin about itself

The statement shows about $1,758.47 a month of living expenses against $1,400 declared. The larger figure was used, and the difference is worth asking about.

The statement the reading was made from

Every figure in this example is invented. There is no borrower, no bank statement and no lender: the transactions below were generated from the numbers in the form so that the policy engine has a statement to read, which is what it reads. 77 transactions between 5 December 2025 and 29 May 2026, which is the 176 days before the day this was assessed. The amounts do not vary from one payment to the next, and real ones do — a variable income is assessed at its lower quartile and a variable expense at its upper one, so a statement this regular gets a friendlier reading than a real one would.

Transactions read
77
Share this software could identify
100%
Period
5 December 2025 to 29 May 2026 — 176 days
Enough history for a fortnightly rhythm
yes
Regular income identified
1 stream
Verified monthly income
$4,799.99
Every one of the 77 invented transactions
DateDescriptionAmount
5 December 2025 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
9 December 2025 BP CONNECT ELSTERNWICK VIC AUS -$138.46
10 December 2025 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $2,215.38
11 December 2025 SPORTSBET DEPOSIT -$69.23
12 December 2025 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
19 December 2025 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
19 December 2025 AGL ENERGY LTD DIRECT DEBIT -$300.00
22 December 2025 RENT HARCOURTS ELSTERNWICK REF 88210 -$1,200.00
23 December 2025 BP CONNECT ELSTERNWICK VIC AUS -$138.46
24 December 2025 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $2,215.38
25 December 2025 SPORTSBET DEPOSIT -$69.23
26 December 2025 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
2 January 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
6 January 2026 BP CONNECT ELSTERNWICK VIC AUS -$138.46
7 January 2026 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $2,215.38
8 January 2026 SPORTSBET DEPOSIT -$69.23
9 January 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
16 January 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
19 January 2026 AGL ENERGY LTD DIRECT DEBIT -$300.00
20 January 2026 BP CONNECT ELSTERNWICK VIC AUS -$138.46
21 January 2026 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $2,215.38
22 January 2026 RENT HARCOURTS ELSTERNWICK REF 88210 -$1,200.00
22 January 2026 SPORTSBET DEPOSIT -$69.23
23 January 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
30 January 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
3 February 2026 BP CONNECT ELSTERNWICK VIC AUS -$138.46
4 February 2026 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $2,215.38
5 February 2026 SPORTSBET DEPOSIT -$69.23
6 February 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
13 February 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
17 February 2026 BP CONNECT ELSTERNWICK VIC AUS -$138.46
18 February 2026 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $2,215.38
18 February 2026 AGL ENERGY LTD DIRECT DEBIT -$300.00
19 February 2026 SPORTSBET DEPOSIT -$69.23
20 February 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
21 February 2026 RENT HARCOURTS ELSTERNWICK REF 88210 -$1,200.00
27 February 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
3 March 2026 BP CONNECT ELSTERNWICK VIC AUS -$138.46
4 March 2026 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $2,215.38
5 March 2026 SPORTSBET DEPOSIT -$69.23
6 March 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
13 March 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
17 March 2026 BP CONNECT ELSTERNWICK VIC AUS -$138.46
18 March 2026 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $2,215.38
19 March 2026 SPORTSBET DEPOSIT -$69.23
20 March 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
21 March 2026 AGL ENERGY LTD DIRECT DEBIT -$300.00
24 March 2026 RENT HARCOURTS ELSTERNWICK REF 88210 -$1,200.00
27 March 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
31 March 2026 BP CONNECT ELSTERNWICK VIC AUS -$138.46
1 April 2026 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $2,215.38
2 April 2026 SPORTSBET DEPOSIT -$69.23
3 April 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
10 April 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
14 April 2026 BP CONNECT ELSTERNWICK VIC AUS -$138.46
15 April 2026 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $2,215.38
16 April 2026 SPORTSBET DEPOSIT -$69.23
17 April 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
20 April 2026 AGL ENERGY LTD DIRECT DEBIT -$300.00
23 April 2026 RENT HARCOURTS ELSTERNWICK REF 88210 -$1,200.00
24 April 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
28 April 2026 BP CONNECT ELSTERNWICK VIC AUS -$138.46
29 April 2026 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $2,215.38
30 April 2026 SPORTSBET DEPOSIT -$69.23
1 May 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
8 May 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
12 May 2026 BP CONNECT ELSTERNWICK VIC AUS -$138.46
13 May 2026 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $2,215.38
14 May 2026 SPORTSBET DEPOSIT -$69.23
15 May 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
21 May 2026 AGL ENERGY LTD DIRECT DEBIT -$300.00
22 May 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69
24 May 2026 RENT HARCOURTS ELSTERNWICK REF 88210 -$1,200.00
26 May 2026 BP CONNECT ELSTERNWICK VIC AUS -$138.46
27 May 2026 DIRECT CREDIT 049211 ACME LOGISTICS PAYROLL $2,215.38
28 May 2026 SPORTSBET DEPOSIT -$69.23
29 May 2026 WOOLWORTHS 2109 ELSTERNWICK VIC AUS CARD 4412 -$207.69

What is kept

Nothing. The application above is in the address bar and nowhere else. There is no account, no cookie, no session and no database on this surface; the reading, the assessment, the two policy runs and the sealed record on the next page were computed to answer this request and do not outlive it. That is a design decision rather than a policy statement: there is no store to write to.

Except the policy. models/policies/example-au.js is committed, and it has to be — the whole product is that the March rules are still exactly the bytes that were hashed in March. A replay against a policy that lived only in a running process would prove nothing.

And the request itself. A web server writes an access log, and the address bar is where this application lives, so a deployment of this surface that logs full URLs is logging the figures somebody typed. That is the operator's decision and it is worth making deliberately.

What this is not

This is a report of which policy conditions were met on the evidence supplied. It is not an approval, a decline, a recommendation or a score. Creditcrest Technologies is not a credit provider, is not a credit assistance provider, and does not hold an Australian credit licence. Its software produces evidence; the licensee makes the decision.

The policy format has no word for an approval. approve, decline, score and recommend are parse errors in it, and the error message says where the responsibility sits. What a run produces is a list of conditions and, beside each one, whether the evidence met it, did not meet it, or was not enough to say.