Latin American fashion as JSON, size by size

Per size. A shoe listed in 12 sizes comes back as 12 rows, each with its own variant code (for example KO992SHF67UFS-53 for size 35), an in_stock flag, the delivery type and the seller who ships it. sizes_in_stock and sizes_total sit next to the list, so on the sample product 5 of 7 sizes were buyable. The size names and which sizes were buyable were checked against the product page itself on 14 products and matched on 14 of 14. This is Brazil only; the Colombian storefront does not publish size names or stock at all, and says so instead of guessing.

Because Dafiti stops counting at 11. Measured across 478 size-variants on 59 products, every quantity it publishes is between 0 and 11 and 250 of them are exactly 11, so 11 means 'eleven or more', not eleven. When that happens you get stock_qty: null, stock_at_least: 11 and stock_capped: true; when Dafiti really does publish an exact figure, 0 through 10, it is in stock_qty. For the same reason stock_total is null whenever any size is capped and stock_total_at_least carries the lower bound - summing a ceiling would invent a number Dafiti never published.

Because Brazilian shops quote two. price_pix_brl is the cash price the storefront headlines, price_card_brl is what the same item costs on a card, and wherever the cash discount applies the card price is exactly 10 percent higher - 144.43 against 129.99 on the sample product. list_price_brl holds the struck-through original (199.99) and discount_pct is computed from the prices, with the site's own badge kept separately in discount_pct_source so you can see if they ever disagree. Where a product has no cash discount the two prices are equal. All three were verified against the product page's own machine-readable values on 20 products with no mismatches.

Yes, and only the ones Dafiti applies are offered. Measured on tenis on 2026-09-23: 34,678 results unfiltered, 857 for Nike, 9,763 for the colour Preto, 14,135 in size 40, 4,615 sold by Dafiti itself, 9,783 with free shipping, 6,743 between 100 and 200 BRL, and combining Nike with Preto gives 219 and adding size 40 gives 104. Five sorts are offered - relevance, price_asc, price_desc, discount and newest - and each returns a different first row and a different price ladder. Dafiti also accepts a popularity order and then ignores it, returning the identical 20 products in the identical positions as the default, so that one is deliberately not offered here. Every search response also carries a facets block listing the brands, colours, sizes and sellers this particular query accepts, spelled the way the filter wants them, so you never have to guess.

You get an empty result with no_match: true, not a page of unrelated products. This matters because Dafiti answers a keyword that matches nothing with a normal-looking page of 48 products it picked for you, and reports a result total for them. A search for a nonsense string returns zero rows here, and a real search is never mistaken for one.

Brazil (br, BRL) and Colombia (co, COP). Those are the two storefronts that still trade. Dafiti is often listed as a four-country marketplace, but as of 2026-09-23 dafiti.com.ar serves a fashion blog with no catalogue and dafiti.cl serves a page announcing the Chilean store's closure, so country=ar and country=cl are rejected with that explanation. The two live storefronts publish different fields and the difference is documented in the table above rather than hidden behind an average.

Yes. Every Brazilian row carries sold_by_dafiti as a boolean, and product and availability return the seller Dafiti's own page credits for the item, such as adidas or Olympikus. Where an item is listed by an individual trader rather than a company, that person's registered personal name is withheld - Dafiti does not show it on the product page either - and you still get sold_by_dafiti: false so you know it is not first-party stock.

No review texts, only an average and a count, and only in Brazil. No seller ratings and no seller ids. No price or stock history. No size-level stock, no size names, no colour and no seller on search rows in Colombia, and no result total or last page there either. Sponsored placements are not hidden: they are returned as normal rows flagged sponsored, 13 of 58 on one tenis search, so you can drop them yourself if a ranking study needs to.

Dafiti API is a ReefAPI endpoint group for latin america's biggest fashion marketplace: size-by-size availability, brand, colour, the discount against the list price and the marketplace seller. It returns live JSON through POST requests under /dafiti/v1.

Yes. ReefAPI starts with 1,000 free credits, no card required. Dafiti calls use the same shared credit balance as every other ReefAPI engine.

No login to Dafiti is needed for the API response. You call ReefAPI with your x-api-key header, and the playground can run live examples before you create a production key.

The page example is captured from a live search call, and production requests fetch live data through ReefAPI rather than a static sample.

Dafiti actions currently cost 1-3 credits per successful call. Failed or blocked calls are free. All APIs draw from one credit pool.

Yes. Connect ReefAPI once through MCP and your assistant can call dafiti actions with the same key, credit pool and JSON envelope used by normal REST requests.