Cloud object storage is not a POSIX disk you mount like a laptop drive. It is a flat namespace of buckets and keys (paths), with HTTP APIs for put/get/list/delete, versioning optional, and lifecycle rules (“delete after 90 days”). Think AWS S3, Google Cloud Storage, Azure Blob, Cloudflare R2 (S3-compatible API).
Vendors sell durability (many nines), availability by region, encryption at rest, and IAM policies on who can read which prefix. A CDN (CloudFront, Cloud CDN) often sits in front for public assets — same object, cached at the edge.
COFFEE-STORAGE ships a dark UI POC (STORAGE1-CORE) that looks like a storage console. In the actual Coffee stack, that package is documented as the Electron “pick a folder” lane plus a bridge file — not the same thing as the server vault API under /api/vault/*. Picking a folder in Storage does not retarget ROOT_DIR.
So: for exam truth, S3 = public cloud object store. For Coffee, COFFEE-STORAGE is a guided HTML storyboard of those consoles — use the POC to practice UX patterns (buckets, capacity meters, file grids) you will see in real products.
Coffee Server Vault is your tree under one root with IPC — philosophically closer to “self-hosted MinIO” than to “Slack’s SaaS.” The SaaS lesson is about delegating durability and IAM to a provider.
Best inside the full Coffee shell; opening the HTML file alone still shows the UI story.