I kept running into the same thing: someone wanted five rows out of a spreadsheet, and the options were all bad. Publish the sheet publicly and hand out a link that leaks the whole thing. Download a CSV and parse it by hand, every single time. Or wire up the Sheets API, which means OAuth plumbing, a refresh token, and a service account for something that is conceptually one GET request. So I built Sheet2JSON. You paste a Google Sheets link and you get structured JSON — the header row becomes the keys, nothing to map. If you sign in, private sheets work too, read with your own Google permission and never stored anywhere but encrypted. The part I'm most excited about is saved endpoints. Pick your columns and filters in the UI, name the recipe, and you get a stable URL that reads current data: curl "https://sheet2json.vercel.app/ap..." \ -H "Authorization: Bearer s2j_your_key" And select / where / sort / limit run server-side, so pulling five rows out of a big tab doesn't mean transferring all of them: curl "https://sheet2json.vercel.app/ap..." A few decisions I'd defend: - No account for public sheets. An API that needs a signup to try is an API nobody tries. There's an anonymous tier under a per-IP limit. - The scope is spreadsheets.readonly. Not Drive. A Drive scope would turn a routine sign-in into a request for broad access to someone's files, which this app has no use for. - A typo in a column name is a 400, not an empty result. Because "no matching rows" and "you misspelled the column" look identical otherwise, and one of those is a bug. It's built with Bun, TanStack Start and Vercel, and the API index at /api/v1 describes itself in JSON if you'd rather read it that way. Would love to hear what you reach for sheets with — that's the part I'm least sure I've optimized for.

Sheet2JSON

Turn any Google Sheet into usable data

Turn any Google Sheet into usable data