Create Item - MemoryLake AI

POST /openapi/memorylake/api/v1/drives/items

One endpoint, two modes — the item_type field selects which:

Folder (item_type: "folder") — creates an empty folder in the Library.
File (item_type: "file") — registers a file whose bytes are already in storage from a completed chunked upload. You must have called Create Upload first and collected every chunk’s ETag.

Pick the tab below that matches the item type you’re creating. Required fields differ by mode.

parent_item_id accepts the alias MY_SPACE (your workspace root). Pass it directly instead of looking up the root ID. You can also pass any folder’s item_id to nest deeper.

Request Body

Create folder
Create file

item_type

string

required

Must be "folder".

parent_item_id

string

required

Parent folder — MY_SPACE or any folder’s item_id.

name

string

required

Folder name, e.g. Q2 Reports.

name_conflict_strategy

string

default:"rename"

Behavior if the name already exists in parent_item_id:

rename (default) — append a numeric suffix (Q2 Reports_1).
deny — fail with 409 DRIVE_ITEM_CONFLICT.

x_attrs

object

Extended attributes — a string → string map for caller-defined metadata.

This call finalizes an upload you opened earlier — it does not transfer bytes. Before you reach here you should have:

Called Create Upload with file_size and received an upload_id + one pre-signed URL per chunk.
PUT every chunk to its URL and saved the ETag header of each response.

item_type

string

required

Must be "file".

parent_item_id

string

required

Parent folder — MY_SPACE or any folder’s item_id.

name

string

required

File name, e.g. report.pdf.

from

object

required

Reference to the completed chunked upload.

Show from

upload_id

string

required

The upload_id returned by Create Upload.

part_etags

array

required

One entry per uploaded chunk. Include every chunk, in order.

Show part_etags item

number

integer

required

Chunk number (1-based). Must match the number from the upload session.

etag

string

required

ETag header value returned when you PUT that chunk.

name_conflict_strategy

string

default:"rename"

Behavior if the name already exists in parent_item_id:

rename (default) — append a numeric suffix (report_1.pdf).
deny — fail with 409 DRIVE_ITEM_CONFLICT.
overwrite — overwrite the existing file’s content in the Library. The existing item_id is preserved.
replace — delete the existing file and create a new one. A new item_id is issued; prior references to the old ID become stale.

overwrite only replaces the file bytes stored in the Library — it does not re-process or re-index any project documents that were previously imported from the file. To propagate new content into a project, call Add Documents again after the overwrite (or use a fresh file).

x_attrs

object

Extended attributes — a string → string map for caller-defined metadata.

curl -X POST 'https://app.memorylake.ai/openapi/memorylake/api/v1/drives/items' \
  -H 'Authorization: Bearer sk_xxxxxx' \
  -H 'Content-Type: application/json' \
  -d '{
    "item_type": "folder",
    "parent_item_id": "MY_SPACE",
    "name": "Q2 Reports"
  }'

Response

The response shape is the same for folders and files.

data

object

Show Created item

item_id

string

ID of the resulting item. With overwrite, this equals the existing file’s item_id (preserved). With replace, rename, or a fresh creation, a new item_id is issued.

name

string

Effective name. Differs from the requested name when rename produced an auto-suffix.

{
  "success": true,
  "data": {
    "item_id": "sc-5c6bf0f82d624a20a6fa4696997bdd46:7d8cf1e93f634b31b5",
    "name": "report.pdf"
  }
}

Always read name from the response. With the default rename strategy, the server may have appended a suffix — the returned name is what you should display and store.

MemoryLake API

​Request Body

​Response

Request Body

Response