> ## Documentation Index
> Fetch the complete documentation index at: https://docs.onebucket.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Scenarios

> Seven patterns for where OneBucket fits, each building on the last

OneBucket puts one S3-compatible endpoint, and a native MCP server, in front of object storage spread across clouds, on-premises and edge. Applications and agents see one bucket. Underneath, policy decides where data lives, when it is replicated, cached or migrated, and when it is left where it is.

The scenarios below are the patterns that come up most often. The first four are infrastructure patterns and build on each other: one endpoint in front of one backend, several backends behind the same endpoint, a policy that decides which backend holds which bucket, and one of those backends placed next to compute. The next three are AI and analytics patterns on top of that foundation. All of them are configuration: a policy and a key, no code written.

| # | Pattern | The situation it fits |
| - | - | - |
| 1 | [Migrate behind a fixed endpoint](https://docs.onebucket.io/scenarios/migration.md) | A cloud exit, a cost move or a consolidation, and every application has the old bucket hard-coded |
| 2 | [Write once, land in several clouds](https://docs.onebucket.io/scenarios/replication.md) | A second copy of critical data outside the primary cloud, without a sync job per bucket |
| 3 | [Placement by policy](https://docs.onebucket.io/scenarios/placement.md) | Certain data must stay in a region or cloud, under your own keys |
| 4 | [Cache next to compute](https://docs.onebucket.io/scenarios/cache-near-compute.md) | GPUs in one cloud, data in another, idle time and egress on every run |
| 5 | [Agents reach data through MCP](https://docs.onebucket.io/scenarios/agents.md) | Agents or RAG over data in several stores, without handing out raw cloud keys |
| 6 | [Fewer tokens: know before fetching](https://docs.onebucket.io/scenarios/token-economics.md) | Agents spend tokens finding data and carry every wrong file through every later turn |
| 7 | [Parquet across clouds](https://docs.onebucket.io/scenarios/parquet.md) | Tabular data in several stores and every analysis starts by exporting and stitching |

## Combining patterns

Several patterns combine naturally:

* **Placement + replication** when the copies must sit in specific jurisdictions.
* **Migration + placement** when the move is a cloud exit for sovereignty reasons.
* **Token economics + agents** whenever an agent is the consumer.

## What stays the same in every scenario

* OneBucket holds no data of its own. Bytes stream through it; the storage underneath is unchanged.
* Backend credentials stay with you, held in the secure vault. Applications and agents get OneBucket keys or OAuth tokens, never raw cloud keys.
* Placement is set per bucket by policy, not per object.
* Every operation is an event, scoped to organization and key, available over the Events MCP.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.