E-commerce pack

Inventory CSV import API

Import per-SKU per-location inventory levels from CSV. Multilingual warehouse and stock-level hints.

low riskinventory_v1
30-second curl
curl -X POST https://api.adaptivmapr.com/v1/uploads \
  -H "Authorization: Bearer $ADAPTIVMAPR_API_KEY" \
  -F "template=inventory_v1" \
  -F "file=@your_data.csv"
→ 6 canonical fields · 0 validated · low risk

Canonical columns

The whole schema, printed as it ships.

Every canonical column, the type each row carries, whether it is required, the field-level validators that fire on commit, and the multilingual header hints the cascade resolves against. This is the shipped definition, not a summary of it.

inventory_v1
fields
6
required
3
validated
0
hints
16
Canonical columnTypeRequiredValidatorsHeader hints the cascade matches
skustringyesmatched on the column name
location_idstringyeslagerentrepôtmagazzinoalmacénwarehouse
on_handnumberyesbestandstockgiacenzaexistencias
reservednumberreserviertréservériservatoreservado
last_counted_atdatematched on the column name
reorder_pointnumbernachbestellpunktpoint de réapprovisionnementpunto di riordino

Read the same definition as JSON at GET /v1/templates/inventory_v1. A hint match resolves on layer 2 — no LLM call, no token spend, just the flat per-map fee. Hover a validator id to see what it checks.

  • 6 canonical fields
  • 3 required
  • 0 validated
  • 16 header hints, 5 languages

Why it exists

Written for the file you actually receive.

The Inventory template is the per-SKU per-location stock-level row. Each row carries a SKU (required), a location_id (required — multi-warehouse is the common case), on_hand (the physical count), an optional reserved (committed but not yet shipped), an optional last_counted_at timestamp, and an optional reorder_point. E-commerce and warehousing teams use this template for daily stock-sync jobs, for cycle-count reconciliation, and for migrations between warehouse-management systems. The composite key is (sku, location_id) — the cascade does not enforce it because primary-key constraints belong downstream, but the dry-run report flags duplicate composites so you can decide whether to sum, take the latest, or reject. The location_id field picks up Lager / entrepôt / magazzino / almacén / warehouse at zero cost.

on_hand is the only required numeric field — reserved and reorder_point are optional because not every warehouse tracks them. last_counted_at is a date, not a datetime, in most sources; the parser accepts both. The template does not derive available_to_promise (on_hand - reserved); that lives downstream so the formula can be customised per workspace.

Migration scenarios & the foreign headers they ship

Migration scenarios for the Inventory template: daily stock-sync jobs from a WMS into a storefront, cycle-count reconciliation against the source of truth, WMS-to-WMS migrations during a fulfilment-provider swap, and multi-warehouse consolidation where regional stock files need to reconcile into one cross-region view. Foreign headers we see often: "SKU / Artikelnummer / Référence / Codice / Location / Lager / Entrepôt / Magazzino / Almacén / Warehouse / Bestand / Stock / Giacenza / Existencias / Reserved / Reserviert / Réservé / Riservato / Reservado / Last Counted / Letzte Zählung / Reorder Point / Nachbestellpunkt / Point de réapprovisionnement / Punto di riordino". The composite-key dry-run report flags duplicate (sku, location_id) pairs — the most common cause of mysterious stock overcounts in homegrown sync jobs.

The cascade

Five layers, and the cheapest one wins.

Layers run in order and stop the moment a column resolves. That is the single biggest cost lever in the system: a column caught on layer 2 never reaches the metered layer 5.

  1. L1Statisticsno LLM

    Auto-accepts a header that past confirmations already resolved the same way, at {minN:100, minRatio:0.95} or {minN:20, minRatio:1.00}.

  2. L2Heuristicno LLM

    Normalises accents, punctuation and whitespace, then compares against the column name, the label, and every registered hint (DE / FR / IT / EN / ES).

  3. L3Fuzzyno LLM

    Token-set ratio plus Levenshtein over the normalised strings. Auto-accepts at 0.80 — it absorbs typos and reordered words.

  4. L4Semanticcheap, cached

    Embedding cosine between the header and the field’s label + hints. Catches the long tail of paraphrases.

  5. L5LLMmetered

    Everything still unresolved goes up in ONE batched, collision-aware call, constrained to this template’s column set so it cannot invent a field.

Try it

One template id, two ways in.

REST for your import pipeline, MCP for your editor. Both run the same cascade and both honour the same schema-only clamp.

REST · POST /v1/uploads

Name the template; the cascade picks up the rest. The canonical definition is read-only at GET /v1/templates/inventory_v1.

bash
curl -X POST https://api.adaptivmapr.com/v1/uploads \
  -H "Authorization: Bearer $ADAPTIVMAPR_API_KEY" \
  -F "template=inventory_v1" \
  -F "file=@your_data.csv"
→ upload created · mappings ready · confirm before commit

MCP · Cursor / Claude Desktop

Drop AdaptivMapr into your editor and call the same cascade as a tool. Schema-only calls leave only column names and up to three clamped sample rows.

mcp
// In Cursor or Claude Desktop with the AdaptivMapr MCP server installed:
adaptivmapr.match_headers({
  template_id: "inventory_v1",
  headers: ["sku", "location_id", "on_hand", "reserved"]
})
schema-only · headers and ≤3 rows, 80 chars each
MCP install instructions

Questions

Inventory CSV import — FAQ

Lot-level inventory needs a separate template — one row per lot per location. The canonical row is SKU-level because most catalogues are non-lotted.
The validator does not reject negatives. Negative on_hand usually means an over-shipped or mis-counted SKU; flagging that is downstream business logic.
Model it as a separate location_id (e.g., "transit_seattle") or fork the template to add an `in_transit` field. The canonical row is per-physical-location.
The template does not enforce freshness. Downstream business logic decides whether a stale count means "trust on_hand" or "block sale until recounted". The canonical row carries the timestamp; interpretation is per workspace.

Ready when you are

Map inventory in production — without shipping raw records.

Schema-only mode leaves only headers and a handful of clamped samples. Add full-data when you need row-level AI, routed in-region under a BAA.

No free tier · $10 prepaid wallet to start · every map draws a small flat fee, deterministic ones included
Inventory CSV import API — AdaptivMapr — AdaptivMapr