P04 The Paperwork Trial Generated invoice images
Tax ID Collision
A compact case around vendor identity and tax calculation conflicts.
Transparency boundary
This page shows the files the model can inspect. It does not show the hidden oracle solution.
Excluded: ground_truth.json, expected_artifacts.json, manual readings, and calibration notes.
Generated Images
Scans and visual attachments visible to the model
Task And Source Files
No oracle solution filesREADME_TASK.md
# Paperwork Trial v3 Generated Image Case 04 This is synthetic benchmark bookkeeping. It is not financial, tax, or legal advice. Audit the scanned document folder and write `audit_result.json` only. Case ID: `P3-GEN-04` Output keys exactly: `case_id, approved_invoice_ids, review_invoice_ids, reject_invoice_ids, ignored_document_ids, total_approved_gross_cents, warnings_by_invoice, evidence, proof_code` Rules: - Treat the files in `scans/` as the source documents. - Ignore documents that are explicitly statements or not invoice payment requests. - Match vendors by visible vendor name and tax ID when available. - `payment_match` requires paid bank rows for the invoice whose summed amount equals the invoice gross. - `vendor_tax_id_conflict` applies when the visible vendor name matches one vendor but the visible tax ID belongs to a different vendor in `vendor_master.csv`. - `tax_rounding_mismatch` applies when invoice tax differs from the expected tax by more than 1 cent. - For this case, expected tax is 19% of net, rounded to the nearest cent. - Approved invoices have no warnings and paid amount equals gross. - Review invoices have warnings but are not reject-level. `vendor_tax_id_conflict` and `tax_rounding_mismatch` are review-level warnings. - Reject only if the invoice is impossible to process, uses an inactive vendor, or lacks a required PO. - Sort all invoice-id arrays ascending. - Allowed warning codes are exactly: `tax_rounding_mismatch`, `vendor_tax_id_conflict`. - `warnings_by_invoice` must include every real invoice ID and sorted warning arrays. - Warning arrays must be flat arrays of lowercase strings, never nested arrays and never prose labels. - `ignored_document_ids` must include visible document IDs from ignored non-invoice documents, not filenames. - `total_approved_gross_cents` is the sum of approved invoice gross totals only. - `evidence` must list the relative source file paths used in stable alphabetical order, including folder prefixes such as `scans/`. - Include the relevant CSV files and every scanned document inspected in `evidence`, including ignored statement scans. - `proof_code = total_approved_gross_cents + sum(numeric parts of all real invoice IDs) + 97 * total_warning_count`.
bank_export.csv
date,description,invoice_id,amount_cents,status 2026-06-12,Orion Field Services,INV-4170,14500,paid 2026-06-12,Orion Field Services,INV-4171,12000,paid
model_prompt.md
You are auditing a synthetic scanned paperwork folder. Read `README_TASK.md`, inspect the files in `scans/`, and use `bank_export.csv`, `vendor_master.csv`, and `purchase_orders.csv`. Write `audit_result.json` only. This is benchmark bookkeeping, not financial, tax, or legal advice. Important: - The scanned images are the source documents. - Ignore statements and documents that are not invoice payment requests. - Do not invent fields that are not supported by the files. - Use visible document IDs for `ignored_document_ids`, not filenames. - Use relative paths with folder prefixes in `evidence`, for example `scans/example.png`. - Use only allowed lowercase warning codes from `README_TASK.md`. - Warning arrays must be flat arrays of strings.
purchase_orders.csv
po_id,vendor_id,limit_cents,status PO-4170,V-912,16000,open PO-4171,V-912,13000,open
vendor_master.csv
vendor_id,name,tax_id,status V-912,Orion Field Services,OF-912,active V-913,Orion Systems,OS-913,active