ifood.com.br logo
ifood.com.br

list-stores-with-ratings

Installation

Adds this website's skill for your agents

 

Summary

Extract a ranked list of iFood stores for a delivery address, with ranking, rating, category, name, region, and store link (review counts are gated behind an Akamai challenge).

FIG. 01
FIG. 02
FIG. 03
FIG. 04
SKILL.md
122 lines

List iFood Stores with Ratings

Purpose

Produce a ranked list of stores (restaurants, markets, etc.) that iFood shows for a given delivery address on www.ifood.com.br/restaurantes. For each store it returns its ranking (list position), rating (nota), category (categoria), store name, region (região), and the canonical store link. This is a read-only browsing task — no login, no ordering. Note that the number of ratings/reviews (número de avaliações) is NOT exposed on the store list and is only visible on each store's detail page, which is gated by an Akamai "Press & Hold" bot challenge (see Gotchas).

When to Use

  • You need a ranked feed of the stores iFood surfaces for a specific address/neighborhood in Brazil, with their star rating and category.
  • You want the canonical /delivery/... store URLs (and the region slug embedded in them) for a set of stores.
  • You are building a leaderboard / comparison of stores by rating in a given region.
  • Do not use this when you need per-store review counts in bulk — that data lives behind a bot wall (documented below).

Workflow

iFood has no usable public listing API for this (the marketplace.ifood.com.br endpoints are Akamai-protected and return 404/challenges to direct fetches), so the recommended method is a browser session with stealth enabled. The single hard requirement is that a delivery address must be set before any store is visible.

  1. Start a stealth session and open the list page. Use a remote Browserbase session with --verified --proxies (Brazilian storefront + intermittent Akamai challenge). Navigate to https://www.ifood.com.br/restaurantes.
  2. Handle the mandatory address modal. The page immediately shows a blocking modal "Onde você quer receber seu pedido?". Click the "Buscar endereço e número" textbox and type a real Brazilian address, e.g. Avenida Paulista, 1578.
  3. Pick a suggestion. After ~2s an autocomplete list renders. Click the first suggestion whose subtitle contains the target city/state, e.g. Bela Vista, São Paulo - SP, Brasil.
  4. Save the address. A confirmation form appears with the street number pre-filled. Click "Salvar endereço". The page reloads with store cards under the "Lojas" heading.
  5. Extract the cards from the DOM (not snapshot). Use browse get markdown body. Each store card is a single link of this exact shape: 
    [<logo-img>{Name}{rating} • {category} • {distance} km](/delivery/{region-slug}/{store-slug}/{merchant-uuid})
    Parse each card into fields:
    • ranking — 1-indexed position in the "Lojas" list.
    • name — text before the first decimal number.
    • rating — the first decimal number (e.g. 4.8).
    • category — the middle •-delimited segment (e.g. Mercado, Brasileira, Árabe).
    • region — the {region-slug} path segment (e.g. sao-paulo-sp).
    • link — https://www.ifood.com.br + the href.
    • review_count — set to null; it is not present on the list card.
  6. Emit JSON matching the schema in Expected Output.

Note on review counts: if número de avaliações is truly required, you must open each store's /delivery/... detail page individually and solve the Akamai Press & Hold challenge there. This is slow and unreliable (the challenge frequently rejects automated holds), so treat review counts as best-effort/null in any bulk pass.

Site-Specific Gotchas

  • Address is mandatory and blocking. Nothing renders until a delivery address is saved. The whole page is a modal on first load. Use a well-known address (e.g. Avenida Paulista, 1578) to get a dense São Paulo store list; the region you pick determines every store and region slug returned.
  • /restaurantes is the ranked storefront. It renders ~450+ store cards under the "Lojas" heading, already ordered by iFood's ranking for that address. Category landing pages /mercados, /bebidas, /farmacia, /petshop behave the same way; /busca?q=<term> is the search variant.
  • Review count is not on the list card. List cards expose only rating, category, and distance. número de avaliações (e.g. "500+ avaliações") lives on the store detail header only.
  • Akamai "Press & Hold" wall on detail pages (confirmed). Opening any /delivery/{region}/{slug}/{uuid} store page triggers an Akamai Bot Manager challenge (nested iframes, eng5outu service-worker path, "Before we continue… Press & Hold to confirm you are a human", with a Reference ID). The merchant content (name/rating/review-count/menu) never loads until it is solved. A synthetic hold via browse mouse drag (same start/end point, --steps 70 --delay 100, ~7s) was rejected with "Please try again", and --verified sessions solve it only intermittently. Don't rely on scraping detail pages in bulk.
  • The list DOM survives the overlay. On the /restaurantes list page the same challenge overlay can appear, but the store cards are already in the DOM behind it — browse get markdown body / browse get text "a[href*='/delivery/']" still return all the card data, so list extraction succeeds even when the overlay is showing.
  • Prefer markdown/text extraction over browse snapshot. The browse CLI prints an "Update available: 0.7.2 -> 0.9.6" banner to stderr that the autobrowse harness treats as a tool error, wasting turns; browse type emits it too (keystrokes still land). Use browse get markdown body for reliable extraction.
  • Direct API is a dead end. marketplace.ifood.com.br/v1/merchants?... returns 404 Not Found, and the host sets Akamai ak_bmsc cookies — confirmed not worth fetching directly. The sitemap (static.ifood.com.br/sitemap/br/sm-0.xml) only lists category/search landing pages, not per-address store lists.
  • Region comes from the URL, not visible text. The region field is the slug in the store link (sao-paulo-sp); it reflects the store's own city/state, which matches the address you set.

Expected Output

Primary (successful list extraction):

{
  "success": true,
  "address": "Avenida Paulista, 1578 - Bela Vista, São Paulo - SP",
  "count": 20,
  "stores": [
    {
      "ranking": 1,
      "name": "Encanto da Uva - Vinhos & Cia 24h",
      "rating": 4.8,
      "review_count": null,
      "category": "Mercado",
      "region": "sao-paulo-sp",
      "distance_km": 3.7,
      "link": "https://www.ifood.com.br/delivery/sao-paulo-sp/encanto-da-uva---vinhos--cia-24h-vila-nova-conceicao/bbce6e34-9d5f-4c42-b830-ac005fdb466e"
    },
    {
      "ranking": 2,
      "name": "Maria's Pamonha",
      "rating": 4.9,
      "review_count": null,
      "category": "Brasileira",
      "region": "sao-paulo-sp",
      "distance_km": 0.8,
      "link": "https://www.ifood.com.br/delivery/sao-paulo-sp/marias-pamonha-bela-vista/42ba93c5-9909-4080-8e3d-b57372d60f6f"
    }
  ],
  "error_reasoning": null
}

Field notes:

  • review_count is always null from the list page (see Gotchas). Include the key for schema stability.
  • distance_km is a bonus field parsed from the card; omit if not needed.

Blocked outcome (challenge prevented reaching the list, e.g. bare session fully walled):

{
  "success": false,
  "address": "Avenida Paulista, 1578 - Bela Vista, São Paulo - SP",
  "count": 0,
  "stores": [],
  "error_reasoning": "Akamai 'Press & Hold' human-verification challenge blocked the page before any store card rendered (Reference ID shown in overlay)."
}