Evaluations

Explore a profit-and-cash workbook built with fictional data. Then reproduce 3 accounting tool evaluations at fixed releases, using fictional inputs and expected results recorded by hand.

An evaluation shows what one version did on declared fixtures. It does not prove completeness, suitability or a professional conclusion.

For identity, source and release assurance, see the Evidence register. Each evaluated tool is listed on the Tools page. Choose an example for any tool.

Last reviewed 26 September 2026.

$35,957.55 profit, yet a $25,160 cash shortfall

Lumbridge Services is a fictional maintenance business in Newcastle, NSW. Moving one $44,000 customer receipt from 14 October to 28 November 2026 leaves quarterly profit before income tax unchanged at $35,957.55 while the daily cash balance falls to −$25,160.00 on 6 November. All amounts are AUD and every figure is fictional.

Summary of the worked example. Scroll sideways on small screens.
Forecast measure	On time	45 days late
Profit before income tax	$35,957.55	$35,957.55
Lowest daily closing cash	$16,800.00 on 7 October	−$25,160.00 on 6 November
December closing cash	$67,320.00	$67,320.00

Read the full worked example: the business, the assumptions, the working behind both cases, the workbook download, reproduction commands and limits.

This is a maintainer-prepared demonstration. Its calculations have technical verification records. No client outcome, forecast accuracy or time saving has been measured.

Choose an example to reproduce

Each example names its inputs, the command or browser steps, the expected result and the review limits. The 3 fixed-release evaluations below retain their original versions. Package trials and source tests are labelled separately so a newer installation is not mistaken for the recorded run.

Examples for each tool. All inputs are fictional. Scroll sideways on small screens.
Tool	Reproduction route	What to check
Xero trial balance CSV export	Fixed-release evaluation	One accepted export and two refusals.
Subcontract ledger skills	Agent evaluation procedure	Declared checks and prohibitions, judged per fresh model session.
Construction WIP schedule	Pinned source test	50% progress and a $50,000 contract asset.
Payday Super timing	Pinned package trial	AT_RISK while fund receipt is missing.
ATO benchmark comparison	Pinned source test	30% cost of sales after excluding the separately mapped wages.
Coal LSL levy calculator	Browser calculator	Choose Load the synthetic example, then Calculate; compare the inputs and formula with the fixed homepage proof.
Trust distribution checks	Pinned package trial	$100,000 on the declared section 99B assumptions, plus a refusal case.
Company tax and franking checks	Pinned package trial	25.0% on the declared base rate entity inputs, plus an unsupported-year refusal.
Workpaper Review Gate	Fixed-release evaluation	NOT_READY findings, then READY after the stated corrections.
Monthly Close Controls	Pinned source case	REVIEW with 8 exceptions in the supplied July close.
Australian tax tools for AI agents	Pinned package demonstration	Two synthetic tool calls, without an agent session or credentials.
Ozzit Excel LAMBDA library	Pinned workbook	Download release v3.4.2, check the published SHA-256 and confirm one LAMBDA result in desktop Excel.

The Excel cash flow case and browser planning calculators offer examples without Python. For an agent evaluation, preserve the source revision, model, runtime, available tools and full response. Repeat the procedure and assess the required checks; identical prose is not expected.

Short accounting cases

Start with a review question and inspect the synthetic inputs, command and output. These are maintainer-prepared examples, not independent firm evaluations.

Balanced trial balance, unfinished review

The synthetic July close returns REVIEW: 8 exceptions, none blocked. Three findings explain why the balanced trial balance still needs attention.

Three findings from the supplied case. The 8 exceptions are not 8 accounting errors.
Finding	Evidence	Reviewer action
Creditors reconciliation	Trial balance less supplied subledger: −$250.	Trace the difference and document its cause.
Account mapping	Operating Expenses account 6000 has no review group mapping.	Confirm the mapping and review owner.
Financial year reset	The comparison crosses from June to July.	Use a suitable period comparison for P&L YTD movements.

Monthly Close Controls guide: inputs, exception report and reproducible command.

Inspect the case, files and reproduction command at the captured source commit. The source records reproduction on 6 September 2026. A person must resolve the evidence before signing off.

Payday Super: sent before the deadline, received when? A remittance record and a fund receipt support different conclusions.
Division 7A: does the repayment count? Inspect a $10,000 change in the repayment assertion without treating the arithmetic as repayment classification.
Month-end close: balanced trial balance, unfinished review. Trace a $250 creditors difference and the June-to-July comparison warning.

Workpaper Review Gate evaluation

This evaluation runs 2 fictional BAS packs through Workpaper Review Gate, fixes the stated failures and reproduces the READY result.

Fictional input: Two BAS workpaper packs.
Expected result: Exact NOT_READY findings, then READY after correction.
Fixed release: Workpaper Review Gate review-ready-gate/v0.1.3.
Limitation: READY is readiness for a manager, not sign-off.

Evaluation source, Workpaper Review Gate evaluation · Primary source: Australian Taxation Office, Business activity statements

Xero trial balance integrity evaluation

This evaluation runs 3 fictional CSV files through the movement and year-to-date balance checks without connecting to Xero.

Fictional input: Balanced and unbalanced trial balance CSVs.
Expected result: One accepted export and 2 explicit refusals.
Fixed release: Xero trial balance export xero-trial-balance-export/v0.1.6.
Limitation: No tenant credentials, live tenant data or API availability test.

Evaluation source, Xero trial balance integrity evaluation · Primary source: Xero, Accounting API Reports

Payday Super timing evidence evaluation

This evaluation reproduces 4 evidence-timing outcomes, including at-risk, on-time and late results.

Fictional input: Four contribution and receipt evidence fixtures.
Expected result: Exact status and due date output for each line.
Fixed release: Payday Super Checker payday-super-checker/v0.1.3.
Limitation: Supplied evidence is classified; missing facts are not inferred.

Evaluation source, Payday Super timing evidence evaluation · Primary source: Superannuation Guarantee (Administration) Act 1992 and Australian Taxation Office, LCR 2026/2