tape-gateway, served on its own listener. Any gateway operator can enable it, and workloads that outgrow a hosted offering can run their own gateway.

An S3 client against the gateway: configure it, put an object, list, get it back.
Key concepts
Writes have one extra rule: the gateway writes to your tape as an on-chain delegate. The tape must delegate to the gateway’s key before any S3 write succeeds (delegation).
How access works
Three levels:- Anonymous requests can read and list, subject to the gateway’s metering (rate limits).
- Signed requests use SigV4 credentials scoped to specific buckets or to your own tape.
- Writes additionally require the on-chain delegation.
Set up a bucket
Every S3 tool needs the same two things: a bucket and a credential. Four steps, once per bucket. For Tapenet, complete the wallet and funding setup before reserving a tape. 1. Reserve a tape. Size it for what you will store, and pick how long it should live.--duration buys enough epochs to cover that time; epochs last one hour on Tapenet (epochs). Add --dry-run to see the cost and your balance first.
Output
Output
max_object_bytes is the largest single upload the gateway accepts. Set the delegate on the tape:
Output
tape revoke.
3. Get a credential. The gateway operator issues an access key pair scoped to your bucket through the gateway’s admin API (configuration). If you run the gateway, that is you.
4. Use the label as the bucket name. tape info prints it again whenever you need it:
Output
my-bucket on every S3 page. The base58 address works too with clients that accept it.
Then pick your tool from the quickstarts.