MercadoLibre Reviews Scraper

Scrapes public product review data from MercadoLibre Mexico (mercadolibre.com.mx) without any login. Useful for market researchers, e-commerce analysts, and competitive-intelligence teams who need per-product review signals from MercadoLibre Mexico while staying strictly on public, logged-out pages.

What it actually scrapes (read this first)

Individual product reviews (review title, review text, review date) on MercadoLibre Mexico are gated for logged-out visitors: product detail pages redirect to an account-verification page, and the public API returns 403 for anonymous requests (evidence below). This actor therefore scrapes the closest reachable public surface: the public deals page (https://www.mercadolibre.com.mx/ofertas), which embeds a per-product aggregate review rating ("Calificacion 4.8 de 5 estrellas") in its page data.

Each output record carries the product's title, URL, and aggregate rating (rounded to an integer on a 1-5 scale). The review_title, review_text, and review_date fields are present in the schema but are always null, because the individual review records they would hold are not publicly reachable.

Inputs

Field	Type	Default	Description
start_urls	array of strings	["https://www.mercadolibre.com.mx/ofertas"]	Public MercadoLibre Mexico listing pages to scrape for per-product review ratings. The default is the public deals page, the only surface verified reachable without login.
max_items	integer	48	Maximum number of product records pushed to the dataset. The default deals page carries ~48 product cards.
country	string	"mx"	MercadoLibre country site. Only "mx" (mercadolibre.com.mx) was verified against the public review surfaces; other values log a warning.

Sample output record

{
  "product_title": "Soldadora Inversora Portatil Mma+tig Maquina 2en1 Color Amarillo",
  "product_url": "https://www.mercadolibre.com.mx/soldadora-inversora-portatil-mmatig-maquina-2en1-color-amarillo/p/MLM63723900",
  "rating": 5,
  "review_title": null,
  "review_text": null,
  "review_date": null
}

rating is parsed from the page's "Calificacion 4.8 de 5 estrellas" text and rounded to an integer (4.8 -> 5). Products whose card has no rating component get rating: null.

Coverage and limits

• ~48 records per run from the default deals page (one page, limit: 48).
• Only aggregate ratings are available publicly; individual review text, titles, and dates are gated (see evidence below).
• Plain HTTPS with a normal browser User-Agent, ~1 request/second between pages, no proxies, no JavaScript rendering.
• Page structure changes on MercadoLibre's side can break the embedded-JSON parsing; the actor logs a warning and returns zero rows for that page rather than failing the run.

Evidence that individual reviews are gated (verified 2026-09-25)

• https://www.mercadolibre.com.mx/<slug>/p/MLM63723900 (real product detail page): HTTP 302 -> https://www.mercadolibre.com.mx/gz/account-verification?go=...
• https://listado.mercadolibre.com.mx/celular-smartphone: HTTP 302 -> https://www.mercadolibre.com.mx/gz/account-verification?go=...
• https://api.mercadolibre.com/items/MLM1040520105: HTTP 403 (PA_UNAUTHORIZED_RESULT_FROM_POLICIES) for anonymous requests; the same for /sites/MLM/search. MercadoLibre retired anonymous access to these endpoints.
• https://www.mercadolibre.com.mx/noindex/catalog/reviews/MLM63723900: HTTP 200 and not redirected, but the review list renders client-side only; a plain fetch shows "Este producto aun no tiene comentarios" / "Algo salio mal" and the underlying XHR endpoint is not discoverable from the public shell.
• https://www.mercadolibre.com.mx/ofertas: HTTP 200, no redirect, embeds per-product rating data. This is the surface the actor uses.

Data rules

• Public pages only. No login flows, no session cookies, no CAPTCHA solving, no proxy services.
• No personal data. Reviewer identities are never collected: there is no reviewer name, username, email, or identifier field in the output, and none is extracted at any stage.

Local test

cd mercadolibre-reviews-scraper
mkdir -p storage/key_value_stores/default
echo '{"start_urls":["https://www.mercadolibre.com.mx/ofertas"],"max_items":48,"country":"mx"}' \
  > storage/key_value_stores/default/INPUT.json
/path/to/venv/bin/python main.py