## Documentation Index

Fetch the complete documentation index at: [/llms.txt](https://developer.mindbridge.ai/llms.txt)

Use this file to discover all available pages before exploring further.

Importing **chunked files** allows you to bypass the 8 GiB import limit by splitting a large file into smaller files (“chunks”). Each chunk can be individually retried in case of a network failure, but the entire import process must be completed within 7 days.

### How it works:

- Create a chunked file.
- Add chunked file parts.
- Create a file manager file from a chunked file.

### Add Chunked File Part
This endpoint is used to append a `chunkedFilePart` to a chunked file.

Chunked file parts must provide an `offset` and a `size` along with content. The first chunked file part must have offset `0`. You may add the same chunked file part multiple times to retry a failed import. When retrying, you must use the same `offset` and `size`. You may add chunked file parts out of order.

- Maximum chunked file part size: 50 MiB.
- Maximum number of chunked file parts in one file: 1000.
- Maximum overall file size: 20 GiB.

The body of the request is an HTTP form with the `Content-Type` header set to `multipart/form-data`, and contains 2 parts: `chunkedFilePart` for the **chunk properties**, and `fileChunk` for the **data**.

The `chunkedFilePart` must have the `Content-Type` header set to `application/json` and is similar to other entity types. It specifies the `offset` and `size` of the chunk.

The fileChunk contains the file to be imported. Its `Content-Type` header must correspond to the file’s data type.

For example, for a `.csv` file, the `Content-Type` header for the `fileChunk` should be `text/csv`.

### Permissions
- `api.file-manager.write`