Keep the purchase. Keep the source.

A working offline archive, using invented records.

This proof turns a CSV into a purchasing book you can open in a browser without a server. Search the purchase order, SKU, or supplier. Every accepted line retains a reference to its input record.

Open the working archive → Download archive, source and tests (ZIP, 18 KB)

Try the actual lookup

In the archive, search 00123. It appears on two purchase orders in different currencies. Select USD to narrow the result, then clear the filters. A nonexistent SKU produces an explicit empty result.

Verified: 22 local tests passed.
Fixture: 3 purchase orders, 2 suppliers; all data is invented.
Coverage: 8 input records = 4 accepted + 3 rejected + 1 duplicate.
Ordered value: USD 95.00 and EUR 12.00, kept separate.

Ordered line value means ordered quantity multiplied by unit cost. It is not actual spend, payment, received value, tax, or inventory valuation. One line has an unknown received quantity; it stays unknown.

What survives the conversion

Leading-zero SKUs, partial receipts, source references, and currency labels survive. Invalid quantities are reported instead of silently dropped. Exact duplicate rows are excluded and counted. A source SHA-256 fingerprints the input file.

The source values are escaped before being displayed as HTML. Formula-like text is not executed. The generated archive has no external scripts, fonts, uploads, or network requests. Download it and keep using it independently of Toledo.

Run the converter

Extract the ZIP. Python 3.10 or later is sufficient; no extra packages are needed.

python3 archive.py fixture.csv --output preview --synthetic
python3 -m unittest -v

ZIP SHA-256

810470db5b9fd5ce19ce0367327ae74a6dba48690117b6c1d0beda1e7a591518

The boundary of this proof

The input uses an explicit demonstration schema. No actual Stocky export has been inspected, so Stocky compatibility is unverified. This is not a live connector, an inventory system, or a way to restore historical purchase orders in Shopify.

Source coverage applies only to the supplied file. It cannot establish that an export contains every purchase, recover absent data, or determine whether identical real lines are legitimate without their source identifiers.

Inspect the synthetic records (JSON) Inspect the test results (JSON)

AI agents built and checked this sample for Toledo Technologies LLC. It is a bounded demonstration, not a client deployment or a completed commercial product.