Find Craft Non-Alcoholic Beer Restaurants

Purpose

Given a latitude/longitude (and optional radius), return the list of bars and restaurants near that point that serve Nonny — a Canadian craft non-alcoholic beer brand — sorted by distance. Each result includes name, full address, distance, phone (when listed), and website (when listed). The skill exposes the same data that powers the public store-locator at https://us.nonny.beer/pages/find-us. Read-only; never submits any form on the site.

When to Use

• A user asks "where can I drink Nonny (or craft NA beer) near me / in {city}".
• An agent comparing NA-beer-on-tap availability across a metro.
• Any flow that needs Nonny's bar/restaurant footprint by city — the JSON API is faster, cheaper, and more reliably structured than scraping the Shopify+Stockist storefront.
• Cross-checking whether a specific restaurant carries Nonny (its name will appear in the response near that lat/lng).

Workflow

The find-us page is a thin Shopify wrapper over the Stockist.co store-locator widget (Nonny's widget tag is u10642, hard-coded in the page HTML at <stockist-store-locator data-stockist-widget-tag="u10642">). The widget's runtime hits a public, unauthenticated JSON API at https://stockist.co/api/v1/u10642/locations/search. Hit the API directly — no cookies, session, stealth, or proxies required (validated unauthenticated, no-proxy, from a non-residential IP).

1. Geocode the user's location to lat/lng

The Stockist API does not accept address strings (an address= param returns 0 results — confirmed). The caller must supply numeric latitude and longitude. Use any geocoder (Google Geocoding, Nominatim, Mapbox, the LLM's own knowledge of metro centroids, etc.) to convert "San Francisco", "Vancouver, BC", a postal code, or a user-supplied address into a (lat, lng) pair.

Reference centroids known to return good results during validation:

• Vancouver, BC: 49.2827, -123.1207 → 170 bars/restaurants within 100 km
• Seattle, WA: 47.6062, -122.3321 → 205 bars/restaurants within 500 km
• San Francisco, CA: 37.7749, -122.4194 → sparse (1 within 50 km — "The New Bar")
• Portland, OR: 45.5239, -122.6760 (the widget's default map center)
• Toronto, ON: 43.6532, -79.3832 → 6 bars/restaurants within 500 km
• New York, NY: 40.7128, -74.0060 → 1 bar/restaurant within 500 km

Nonny's footprint is dense in BC and the US Pacific Northwest, sparse east of the Rockies. Set caller expectations accordingly.

2. Call the Stockist locations/search API

GET https://stockist.co/api/v1/u10642/locations/search
    ?latitude={lat}
    &longitude={lon}
    &distance={radius_km}     # optional; default behaviour caps at 500 locations total

Response is application/json shaped:

{
  "locations": [
    {
      "id": 125224191,
      "name": "The New Bar",
      "latitude": "37.79696193",
      "longitude": "-122.43502006",
      "address_line_1": "2181 Union Street",
      "address_line_2": "UNIT A",
      "city": "San Francisco",
      "state": "CA",
      "postal_code": "94123",
      "country": "United States",
      "phone": "",
      "website": "https://thenewbar.com/",
      "email": "",
      "description": "",
      "filters": [{ "id": 11202, "name": "Bars / Restaurants", "position": 1 }],
      "custom_fields": [],
      "distance": 2.8,
      "distance_units": "km"
    }
  ]
}

Pre-sorted by ascending distance. No auth, no headers, no Referer needed. No rate limits observed during testing.

3. Filter client-side to Bars / Restaurants

The response includes four kinds of stockists, identified by the filters[].id (or filters[].name):

Server-side filtering by tag=, tag_filter=, or filter[]= is silently ignored — the server returns the unfiltered set regardless. Always filter in your own code:

const bars = data.locations.filter(loc =>
  loc.filters.some(f => f.id === 11202)
  // or: f.name === "Bars / Restaurants"
);

A single location can carry multiple filters (e.g. "The American" in Vancouver is both Bars / Restaurants and Grocery) — use .some(), not strict equality on the array.

4. Project the fields you need

For the task ("best restaurants in my area with craft NA beer"), the useful subset per result:

• name
• address_line_1, address_line_2, city, state, postal_code, country (concatenate for a display address)
• latitude, longitude (for mapping / re-distancing)
• distance + distance_units (always reported as km, even when you pass units=mi)
• phone, website, email — frequently empty strings; treat as optional
• filters[] — useful if you want to flag dual-category venues ("also sells bottles to-go")

"Best" is not represented in the Stockist payload — there is no rating, review, or priority score that varies between bars (priority: 0 for every Bars/Restaurants entry observed). If the caller wants ranking beyond distance, layer in a secondary signal (Google Places rating, Yelp stars, etc.) keyed by name + address or (latitude, longitude). Without that, sort by distance ascending and return the top N as a reasonable proxy for "closest = best".

Browser fallback

When the Stockist API is unreachable (network egress restrictions, an outage), the user-facing UI at https://us.nonny.beer/pages/find-us reproduces the same data via the embedded widget:

1. browse open https://us.nonny.beer/pages/find-us --remote — no stealth or proxies required; the Shopify host is Cloudflare-fronted and tolerant.
2. browse wait timeout 4000 — the Stockist widget mounts asynchronously after page load.
3. browse click the "Bars / Restaurants" checkbox (accessibility label Bars / Restaurants, near top of the locator region).
4. browse fill the search combobox (placeholder Type a postcode or address...) with the user's city or postal code and browse press Enter.
5. The widget geocodes the input client-side via Google Maps, then hits the same Stockist API. Read browse get text "[role='region'][aria-label='Store locator results']" — it returns a 131 results found... summary plus a flat list of Name / distance / address / website / filters / Directions / View on map per result.
6. For the Canadian storefront (nonny.beer/en-ca/pages/find-us) the same widget tag (u10642) backs the locator — no difference in data, just locale chrome.

The browser path costs ~5-10 turns vs. the API's single HTTP GET. Use it only when the API path actually fails.

Site-Specific Gotchas

• Widget tag u10642 is the only Nonny-specific identifier. It is embedded in the page HTML and is part of every API URL. There is no auth key, project id, or token to manage.
• address= parameter does NOT geocode. Sending address=San+Francisco returns {"locations":[]}. The widget geocodes client-side via Google Maps before hitting the API. You must geocode out-of-band.
• Server-side category filters are silently dropped. tag=11202, tag_filter=11202, filter[]=11202 all return the unfiltered set. Always filter client-side on filters[].id or filters[].name.
• Unfiltered /locations/search?latitude=…&longitude=… (no distance) caps at 500 results. Pass distance={km} when you want a known radius; pass a large value (e.g. distance=500) when sweeping a sparse metro.
• distance_units always reports km regardless of input. Passing units=mi does narrow the radius input proportionally, but the response numbers and the label remain kilometres. Do unit conversion client-side if your UI shows miles.
• Coverage is regional. Nonny is BC-based; the database is densest in BC + the US Pacific Northwest. Major US East-Coast and Central metros (NYC, Boston, Chicago, Atlanta, Toronto) return single-digit counts within 500 km radii. If a caller asks for a sparse metro, return what's there and surface a note — don't infer the API is broken.
• priority is always 0 for Bars / Restaurants entries observed. There is no merchant-supplied ranking signal — distance-ordering is the only built-in sort.
• One location can carry multiple filter buckets. E.g. "The American" (Vancouver, 926 Main St) carries both Bars / Restaurants AND Grocery. Use .some() membership tests, not strict scalar equality, when filtering.
• Storefront is Shopify behind Cloudflare. nonny.beer (bare) 302-redirects to a localised subdomain — us.nonny.beer for US visitors, nonny.beer/en-ca/... for Canadian. The Stockist widget is identical on both; you do not need to pick the "right" storefront to get the right data.
• No Referer header gating. The Stockist API accepts requests with no Referer and no Origin — works from curl, the sandbox proxy fetch, or any HTTP client. Don't bother spoofing browser headers.
• Duplicate entries exist. Several restaurants appear twice in the same metro under slightly different name spellings (e.g. "Tacofino Yaletown" and "Tacofino - Yaletown"). De-dupe on (name.toLowerCase().replace(/[\W_]+/g,''), address_line_1.toLowerCase()) if you need a clean set.
• custom_fields is always [] on this widget tag — no extra structured metadata to project.
• No pagination. The API returns every match within the radius in a single response (subject to the 500 cap when no distance is passed).

Expected Output

Successful call — restaurants within radius:

{
  "success": true,
  "query": {
    "latitude": 49.2827,
    "longitude": -123.1207,
    "distance_km": 25,
    "filter": "Bars / Restaurants"
  },
  "result_count": 131,
  "restaurants": [
    {
      "name": "Rogue Kitchen & Wetbar",
      "address": "602 W Broadway, Vancouver, BC V5Z 1G2, Canada",
      "address_components": {
        "line1": "602 W Broadway",
        "line2": "",
        "city": "Vancouver",
        "state": "BC",
        "postal_code": "V5Z 1G2",
        "country": "Canada"
      },
      "latitude": 49.262,
      "longitude": -123.117,
      "distance_km": 0.7,
      "phone": "",
      "website": "",
      "filters": ["Bars / Restaurants"],
      "stockist_id": 125224191
    },
    {
      "name": "Cactus Club Cafe",
      "address": "575 West Broadway, Vancouver, BC V5Z 1E6, Canada",
      "address_components": {
        "line1": "575 West Broadway",
        "line2": "Broadway + Ash",
        "city": "Vancouver",
        "state": "BC",
        "postal_code": "V5Z 1E6",
        "country": "Canada"
      },
      "latitude": 49.263,
      "longitude": -123.114,
      "distance_km": 0.8,
      "phone": "",
      "website": "cactusclubcafe.com",
      "filters": ["Bars / Restaurants"],
      "stockist_id": 125223781
    }
  ]
}

Successful call — sparse metro with zero matches in radius:

{
  "success": true,
  "query": {
    "latitude": 40.7128,
    "longitude": -74.0060,
    "distance_km": 50,
    "filter": "Bars / Restaurants"
  },
  "result_count": 0,
  "restaurants": [],
  "note": "No Nonny stockists tagged 'Bars / Restaurants' within 50 km of New York City. Nonny coverage is concentrated in British Columbia and the US Pacific Northwest. Widen the radius (try 500 km) or expect single-digit results east of the Rockies."
}

Failure — caller did not provide / could not geocode a location:

{
  "success": false,
  "reason": "no_location",
  "error_reasoning": "Stockist API requires numeric latitude and longitude; the 'address' query parameter is silently ignored. Geocode the user's area to (lat, lng) before calling the skill."
}

Failure — Stockist API unreachable (network or proxy issue):

{
  "success": false,
  "reason": "api_unreachable",
  "error_reasoning": "GET https://stockist.co/api/v1/u10642/locations/search returned <status/error>. Retry with backoff, or fall back to the browser flow on https://us.nonny.beer/pages/find-us."
}