Primat API

# Real data, no account (25 results/call, 250 calls/day):
curl 'https://primat.nu/api/v3/demo/products?q=banan'

# Which stores serve a place? The site's own resolver:
curl 'https://primat.nu/api/v3/demo/stores/resolve?place=Uppsala'

# An API key in one call — no email round-trip to start:
curl -X POST https://primat.nu/api/v3/signup \
  -H 'Content-Type: application/json' \
  -d '{"email": "you@example.com"}'

The key is provisional (7-day TTL, tight quota). One click on the confirmation mail (POST /api/v3/confirm/request) upgrades it to the free tier. Machine-readable contract: https://primat.nu/api/v3/openapi.json — self-describing root: /api/v3.

Authorization: Bearer <key> or X-API-Key: <key>. Two key types: secret (primat_live_…, server-side) and publishable (primat_pk_…, safe to embed in client-side apps: locked to your web origins, read-only, own quota, CORS enabled). Manage keys at primat.nu/konto.

• GET /api/v3 no key Self-describing API directory.
• GET /api/v3/openapi.json no key This document.
• GET /api/v3/demo/products no key Try a real product search with NO key: up to 25 results/call, 250 requests/day per IP. Optional stores= scoping, same syntax as the keyed tier.
• GET /api/v3/demo/stores/resolve no key The store resolver without a key (same demo budget).
• POST /api/v3/signup no key Instant self-serve signup: a provisional secret key, no email round-trip. Body: {"email": "you@example.com"}. Returns 201 with the key (shown once). The key is provisional: 7-day TTL, 60 req/min, 2,000 rows/day plus a shared global pool for all unconfirmed keys. Upgrade via /api/v3/confirm/request. 409 if the address already has an established account.
• POST /api/v3/confirm/request Email the one-click link that upgrades a provisional key to the free tier. Authenticate with the provisional key. Rate-limited (3/day per key). The mailed link is valid 24h; clicking it clears the TTL and moves the key to free-tier quotas (60/min, 50,000 rows/day).
• GET /api/v3/me Account, plan limits (from the entitlement matrix) and usage today for the presented key.
• GET /api/v3/stores Store register with coverage tier and confirmed_at.
• GET /api/v3/stores/resolve Place -> ranked nearby stores + the site's own default store selection. Give ONE of: place= (locality name or 5-digit postcode), postcode=, or lat=&lon=. The default_selection is exactly what the consumer site would pick for that visitor — use it directly as the stores= parameter.
• GET /api/v3/products Ranked product search. With stores=: every carrying (store, product) row in your stores. Without: one row per product at its cheapest carrying store nationally. For FULL store catalogs use /api/v3/prices (keyset pages) — search is ranked and bounded, it does not paginate.
• GET /api/v3/products/{chain}/{store_id}/{product_id} One product at one store.
• GET /api/v3/prices Full store catalogs as keyset cursor pages. The bulk surface: every current price row for up to 15 stores, paged with an opaque keyset cursor over (product_id, chain, store_id) — pages are stable, never skip or duplicate a row, and interleave the same product across your stores for easy comparison. Follow next_cursor until null.
• GET /api/v3/deals Current price drops with prices.previous and discount_ratio.
• GET /api/v3/cheap Cheapest products by comparison (unit) price.
• POST /api/v3/batch Up to 100 product lookups in one call (free tier and up; response bounded by your plan's per-call row cap). Body: {"lookups": [{"gtin": "7310865093530"} | {"chain": "willys", "store_id": "2103", "product_id": "100302995_ST"} | {"chain": "willys", "product_id": "100302995_ST"}], "stores": "willys:2103,..." (optional scope)}. GTIN lookups return every matching chain; without a store scope each chain answers with its cheapest carrying store.
• GET /api/v3/history/{chain}/{store_id}/{product_id} Daily price-change history for one product at one store. Paid tiers — Pro: generation-pinned 12 months. Byrå: full history (since Dec 2025). Deep reads are pinned to a committed history generation and time-budgeted (504 history_timeout on overruns).
• GET /api/v3/history/{gtin} History across carrying stores by GTIN/EAN (up to 6 series). Default: each chain's cheapest carrying store. Scope explicitly with stores= (max 6).

Price FACTS. regular: what everyone pays (public campaigns included). member: only set while a member price runs. multiprice/member_multiprice: PER-UNIT price when buying `quantity` (7 with quantity 5 = '5 for 35 kr'). comparison: unit price per kg/l/st computed on the REGULAR price. offer: the active campaign behind the current pricing, null when none. effective: least price a card-carrying shopper pays right now (a convenience fold; combine the facts yourself for anything serious). previous: on /deals results.

Fields: regular, member, multiprice, member_multiprice, comparison, offer, effective, previous. Timestamps: changed_at = when the price last CHANGED, confirmed_at = when the store was last checked. No change since changed_at means confirmed unchanged, not unmeasured.

Every non-2xx answer is {"error": {"code", "message"}} with honest Retry-After/WWW-Authenticate headers. Codes: invalid_api_key, rate_limited, daily_row_budget_exceeded, demo_limit_reached, query_required, bad_request, not_found, product_not_found, place_not_found, upgrade_required, origin_not_allowed, read_only_key, email_in_use, invalid_email, signup_limit_reached, history_timeout, service_unavailable.

Questions, bulk exports (CSV/Parquet, Byrå plan), or something missing? kontakt@primat.nu.