API

STAC 1.0, no dialect

If it works against another STAC catalogue it works here. The extensions we implement are listed in the landing page conformance array.

$ curl -s 'https://api.mirnl.13381338.xyz/search' -d @query.json

{"type":"FeatureCollection","numberMatched":1240,

"features":[{"id":"S2B_32UNB_20260802_L2A",

"assets":{"B08":{"href":"https://data…/B08.tif"}}}]}

200 OK · 84 ms

Endpoints

GET/Landing page with conformance classes

GET/collectionsList collections

GET/collections/{id}Collection metadata and extent

GET/collections/{id}/itemsItems in a collection, paginated

GET/collections/{id}/items/{item}One item with all assets

POST/searchCross-collection search with CQL2 filters

POST/ordersBulk order: a delivery manifest for many assets

GET/orders/{id}Order status and signed download URLs

POST/subscriptionsPush new matching scenes to a webhook

GET/queryablesFields available to CQL2 filters

Examples

query.json

{
  "collections": ["sentinel-2-l2a"],
  "bbox": [8.1, 49.2, 9.4, 50.1],
  "datetime": "2026-04-01/2026-09-30",
  "filter": {"op": "<=", "args": [
      {"property": "eo:cloud_cover"}, 20]},
  "limit": 500
}

bash

curl -X POST https://api.mirnl.13381338.xyz/orders \
  -H "Authorization: Bearer $TG_TOKEN" \
  -d '{"search": {"collections":["sentinel-2-l2a"],
                  "bbox":[8.1,49.2,9.4,50.1],
                  "datetime":"2026-04-01/2026-09-30"},
       "assets":["B04","B08","SCL"],
       "delivery":"manifest"}'

# → 1,240 items · 4.1 TB · manifest ready in ~30s

python

from pystac_client import Client

cat = Client.open(&#x27;https://api.mirnl.13381338.xyz')
search = cat.search(
    collections=[&#x27;sentinel-2-l2a'],
    bbox=[8.1, 49.2, 9.4, 50.1],
    datetime=&#x27;2026-04-01/2026-09-30',
    query={&#x27;eo:cloud_cover': {'lt': 20}},
)
print(search.matched(), &#x27;scenes')

Bulk orders are the polite way to move terabytes. An order returns a manifest of signed URLs plus a checksum file, so you can hand it to aria2c or rclone and saturate whatever line you have. Ad-hoc parallel scraping of the item endpoints gets rate limited; orders do not.