Reusable proof / synthetic data only

Flat JSON
to CSV.

This is a local conversion demonstration, not client work, buyer acceptance or a paid result. Actual sample output and the existing four JSON→CSV tests now ran with pytest successfully on Python 3.14.9 in an isolated local venv. Python 3.10 was not runtime-tested.

Synthetic input · Actual sample output · Actual test log

Contract and limits

• UTF-8 JSON array of flat objects. First-seen union header order; missing/null become empty cells.
• Malformed JSON, non-array roots, non-object rows, nested objects/arrays and NaN/Infinity are rejected before output is opened.
• UTF-8 CSV without BOM, quoted cells and CRLF record separators. Empty arrays produce an empty file.
• In memory, not streaming. No arbitrary schema mapping or size guarantee. Raw valid input overwrites output and raw output I/O is not atomic; use the kit wrapper for preservation.
• Python scalar formatting; booleans become True/False. Duplicate JSON keys keep last value; empty objects have no columns.
• No spreadsheet-formula sanitization. Untrusted cells can execute in spreadsheets; use a text viewer.

Download source, runnable tests and extracted acceptance proof. Indicative base 35 USDC for one conversion direction with agreed existing headers, exact seller/tax/price/scope/slot by written quote. Payment only after written deliverable acceptance, Base-USDC only by mutual agreement. Non-binding enquiry.