Adding Data in Batches

Use the upload URL flow when ingesting large datasets. The flow has two steps:

1. Request an upload URL
2. Upload a JSONL file to that URL

This pattern is ideal for migrations, historical backfills, or scheduled bulk loads that are too large for request/response ingest patterns.

Step 1: Request an upload URL

Endpoint: POST /v1/data/{namespace}/requests/upload-url with an empty object as the request data.

curl -X POST "https://api.safetykit.com/v1/data/products/requests/upload-url" \
  -H "Authorization: Bearer sk_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{}'

Example response:

{
  "status": "pending_upload",
  "requestId": "req_01h2m7qdmdjckc30e1mnq6xqfd",
  "objectKey": "TEAM_ID/data-api-request-input/products/req_01...jsonl",
  "upload_url": "https://...",
  "expires_at": "2026-02-19T12:34:56.000Z"
}

Step 2: Upload JSONL

Upload your data to upload_url with PUT:

curl -X PUT "{upload_url}" \
  -H "Content-Type: application/json" \
  --data-binary @data.jsonl

The PUT {upload_url} call is made to the pre-signed object storage URL returned by the upload URL endpoint above, not to a standalone SafetyKit API route.

JSONL file requirements

• one valid JSON object per line
• every line includes an id

Example data.jsonl:

{"id":"product_1","title":"Item 1","price":24.0}
{"id":"product_2","title":"Item 2","price":18.5}
{"id":"product_3","title":"Item 3","price":9.99}

After upload, SafetyKit processes objects asynchronously and emits results per object. Large requests can take longer to complete, so you should always track the returned requestId and monitor completion using webhooks or polling.

See Request Status and Polling and Receiving Results for retrieval patterns.