PUB CRAWL SERVICE — instructions for an AI assistant
=====================================================

When the user asks you to save, share or map a pub crawl / food crawl, produce a
crawl as JSON in exactly this shape (stops in the order they will be visited):

{
  "title": "Mission Dive Bar Crawl",
  "slug": "mission-dive-crawl",
  "date": "Thu Oct 1, 2026",
  "author": "Pete",
  "stops": [
    {
      "name": "Zeitgeist",
      "address": "199 Valencia St, San Francisco, CA",
      "notes": "Thu 2 PM - 12 AM, beer garden, cheap pitchers",
      "lat": 37.770027,
      "lon": -122.422113
    }
  ]
}

Rules:
- "title" and "stops" are required; each stop needs "name" and "address".
- "address" must be a full street address including city and state/country.
- "notes" is a short line: hours for that day, what to order, why it's on the list.
- "lat"/"lon" are optional. Only include them if you are confident they are
  accurate; otherwise omit them and the service will geocode the address.
- "slug" is optional (lowercase-with-hyphens); it is derived from the title if omitted.
- "date" and "author" are optional free-form display strings.
- At most 50 stops.

How to save it:
- If you can make HTTP requests:
    POST https://crawl.peteftw.com/api/crawls               (create)
    POST https://crawl.peteftw.com/api/crawls/<slug>/stops  (append one stop object to a crawl you created)
  with headers "Content-Type: application/json" and
  "Authorization: Bearer <token>" (ask the user for the token; never invent one).
  Tokens start with "pc_". Copy the token exactly as the user gave it; do not
  retype it from memory. Easiest is to put it in a shell variable once and
  reuse it, e.g.
    TOKEN='pc_...'
    curl -sS -X POST https://crawl.peteftw.com/api/crawls \
      -H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
      -d @crawl.json
  A 401 response explains why (no token, wrong scheme, or token not recognized,
  with the first characters and length of what was received).
  If the slug is already taken a free one is chosen (e.g. "-2" appended), so always
  use the "slug", "map" and "geojson" values from the response — give the links to the user.
  Replacing or deleting an existing crawl needs an admin token (PUT/DELETE
  https://crawl.peteftw.com/api/crawls/<slug>); a 403 means your token can't do that — tell the user.
  Full API schema: https://crawl.peteftw.com/api/openapi.json (human-readable: https://crawl.peteftw.com/api/docs)
- If you cannot make HTTP requests: output only the JSON in a single code block and
  tell the user to paste it at https://crawl.peteftw.com/new

Once saved, the crawl is viewable at:
  https://crawl.peteftw.com/crawl/<slug>-map.html   (interactive map)
  https://crawl.peteftw.com/crawl/<slug>.geojson    (GeoJSON)