For developers and AI assistants
Who's trading where, as data.
Which traders work which London markets, on which days, with hours, pitch and what they sell. Every fact carries where it came from and when we last checked. Reading takes an API key, issued in a minute and free while the plans are worked out.
GET /api/v1/trading_sessions/today
Authorization: Bearer sd_…
{
"trades_on": "2026-11-16",
"starts_at": "2026-11-16T09:00:00Z",
"ends_at": "2026-11-16T17:00:00Z",
"source": "operator_feed",
"vendor": { "name": "…",
"cuisine": "…",
"hygiene_rating": 5 },
"market": { "name": "…",
"area": "…" },
"pitch": { "ref": "14" }
}
- Markets
- 441
- Traders
- 3411
- Trading days, next 14 days
- 24,237
Live counts from the dataset. Coverage and freshness are published at /data.
Three ways in
The same records, in the form your code, your assistant or your crawler reads.
JSON over HTTPS
Sessions by day, a market's roster for a date, a trader with their schedule and catalogue.
Add ?detail=true for the provenance of every field.
- GET /api/v1/trading_sessions/today
- GET /api/v1/trading_sessions/this_weekend
- GET /api/v1/markets/{slug}/roster?date=…
- GET /api/v1/vendors/{slug}
Five read-only tools
POST /mcp, streamable HTTP, your key as a bearer token. Connect it from any MCP
client that takes a header and ask in words: who is trading, where, when. Nothing a tool
call can do changes, creates or deletes anything.
- find_trading_sessions
- get_market_roster
- search_markets
- get_vendor
- dataset_status
Readable without an API
An llms.txt that says what we are and where to look, sitemaps that say what changed and when, and schema.org on every page: an Event per market day, a Dataset, the questions and answers.
A free key today. Paid plans are coming.
Reading takes a key, and a key is free while the plans are worked out: attributed, non-commercial use, with a limit per key and every call on your dashboard. Paid plans will cover what that can't: commercial use, volume, bulk access and someone to talk to.
Read it
- A key, issued in a minute, for the API and the MCP server
- 1,000 calls an hour per key
- Every call on your dashboard: what, when, how fast
- Attribution, under the data licence
- Non-commercial use
Build on it
- Commercial use, licensed
- Higher limits per key
- Bulk access: a dump rather than a crawl
- Someone to talk to when a field changes
Not priced yet. What early interest asks for shapes them.
Register interestQuestions developers ask
Do I need an API key?
Yes. Create a developer account, issue a key, and send it as Authorization: Bearer <key> on every call to the API or the MCP server. Keys are free while the plans are worked out, each one is limited to 1,000 calls an hour, and the data licence asks for attribution.
Can AI assistants use this data?
Yes, with a key. The pages, the sitemaps, llms.txt at https://www.stallday.co.uk/llms.txt and the schema.org markup are open to crawl. The API and the MCP server at https://www.stallday.co.uk/mcp take a key, so what an assistant reads through them is recorded against the developer who connected it. We ask one thing: attribute Stallday.
How do I connect an MCP client?
Point it at https://www.stallday.co.uk/mcp, streamable HTTP, with your key as a bearer token in the Authorization header; a client that lets you set a header takes it there. It serves five read-only tools, find_trading_sessions, get_market_roster, search_markets, get_vendor and dataset_status. Nothing a tool call can do changes, creates or deletes anything.
What does it cost?
Nothing today: a key is free while the plans are worked out, for attributed, non-commercial use. Paid plans are planned for commercial use, higher limits, bulk access and support. They are not priced yet; register a free key at https://www.stallday.co.uk/for-developers, tell us what you are building, and we will write when they open.
What do you record when I call?
Against your key: the endpoint or tool, the path, the status, how long it took and when. No IP address, and no request or response body. It is what your dashboard shows, and it is deleted after 90 days.
Where does the data come from, and how current is it?
From market operators' rosters, from traders who manage their own listings, and from what markets publish about themselves. Every record carries last_verified_at, the date we last checked it, and a GET with ?detail=true returns the source, method and date behind each field. Coverage and freshness are published live at https://www.stallday.co.uk/data.
Can I download the whole dataset?
Ask. Register at https://www.stallday.co.uk/for-developers and say what you need: we would rather hand you a dump than have you crawl us. Bulk access and redistribution are part of the paid plans.
What may I do with it?
The free tier is for reasonable, attributed, non-commercial use; the terms are at https://www.stallday.co.uk/data/licence. Commercial use, bulk extraction and redistribution need a licence, which is what the paid plans are for.
Building something on this?
Issue a key and read. Then tell us what you're building and how you'd read the data: we'll write when the paid plans open, and what you ask for shapes what they include.