Skip to main content
The tape object subcommands work with S3-style objects in a tape bucket, from the terminal. A bucket is a tape, and an object is a track. The objects architecture page covers that model.

put

Uploads a file under a name. The content type is inferred from the file; --content-type overrides it.
Output
Like tape write, it also accepts --message for inline text or --stdin for pipes.

get

Downloads an object by name. The destination defaults to the object’s filename; pass a path, or - for stdout. --gateway fetches through a gateway, and --bucket reads from a bucket address other than the active tape.
Output
Fetches an object’s metadata without downloading its content.
Output

sync

Uploads a directory as objects, one per file, listed by the path relative to the directory. Content types come from the file extensions. Files that have not changed since the last sync are skipped, so a redeploy pays only for what changed. --prefix puts the whole tree under a name prefix, and --force uploads every file.
Output
For arbitrary directory-to-bucket uploads, use this command. For a website, prefer tape deploy <index.html>, which validates the site tree, protects the entry-point publication order, and supports safe post-publication pruning.

ls

Lists the bucket’s objects. A prefix filters the listing, and --delimiter groups common prefixes for directory-style browsing. --limit and --cursor page through large buckets.
Output

rm

Deletes an object by name. This is a real on-chain delete: the track comes off the tape and its capacity is freed (deletion semantics).