> ## Documentation Index
> Fetch the complete documentation index at: https://docs.withsplendor.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Datasets & schema

> Datasets hold your ingested records and expose a queryable schema and readiness state.

A dataset is a named collection of records within a tenant. You ingest data into a dataset, and you query a dataset by its `dataset_id`. Search runs over one or more datasets at a time.

## Listing datasets

`GET /v1/datasets` returns the datasets in the selected tenant. Each entry includes its identifier and current state, which is what you pass to search and export.

```bash theme={null}
curl https://api.withsplendor.com/v1/datasets \
  -H "Authorization: Bearer $SPLENDOR_TOKEN" \
  -H "X-Splendor-Tenant-Id: $SPLENDOR_TENANT_ID"
```

## Schema

`GET /v1/datasets/{dataset_id}/schema` describes the fields Splendor has discovered in a dataset. Splendor infers the schema from the records it ingests, so the fields you can filter, aggregate, and project in [SQL search](/concepts/search-model) reflect what is actually present in your data.

Use the schema to learn which fields exist before building filters or SQL projections against them.

## Readiness

Ingestion and indexing happen asynchronously, so a dataset becomes queryable in stages. Readiness reports which capabilities are available yet:

* **Schema** — the dataset's fields have been inferred.
* **Text search** — records are indexed and answer `text` and `sql` queries.
* **Semantic search** — vector embeddings are built and answer `semantic` queries.

Check one dataset with `GET /v1/datasets/{dataset_id}/readiness`, or check many at once with `POST /v1/datasets/readiness`.

<Tip>
  Poll readiness after ingesting new data. Semantic search becomes available after text search, because embeddings are built once records are indexed.
</Tip>

## Tracking ingestion

Each ingest produces a run you can follow. `GET /v1/ingest/runs` lists runs across your sources and datasets, and `GET /v1/ingest/runs/{ingest_run_id}` reports the progress and readiness of a single run. Use these to confirm that a load finished before you query it.

For how data gets into a dataset in the first place, see [Ingest a dataset](/guides/ingest-a-dataset), [Stream logs](/guides/stream-logs), and [Connect a source](/guides/connect-a-source).