Storage
Manage the storage behind your ecosystems — files, buckets, and data, with access in clear view.
Manage storage in one place.
Storage is the durable layer under everything you build: uploaded files, structured data, and the buckets that decide which of it a given caller can see — all addressable over the API.
Bytes are easy. Access is not.
Sharing data usually costs you something: a copy that drifts from the original, or the entire table handed over because the grant had no smaller shape. A bucket type is a reference to data the ecosystem already owns, never a copy of it — and how much it exposes is yours to choose, so a grant can be as small as the rows your application wrote through it.
What you store, and who may reach it.
Files that never touch the server
uploads and downloads go straight to object storage over short-lived signed URLs; the platform brokers the URL, owns the metadata row, and checks the declared size against your quota before any bytes move.
A grant on the way in, a mode on the way out
every read is resolved against a grant at the moment it is made. What a delete removes depends on the type: the row itself from a whole-table type, which is the default, and only the reference from one that lists rows.
One layer, two callers
an agent’s tool call re-enters the same route an application’s REST call does, so both are authorized the same way.
Scope it, fill it, share it.
- Open Storage inside your ecosystem, where buckets, access, and all data sit together.
- Add the bucket types the product needs; upload files through the storage API, which hands back the signed URL to put the bytes on.
- Attach an access list where the grant belongs — it covers that bucket and every bucket nested beneath it.