forget

Forget data: one item, a dataset, its memory only, or everything

Remove data from the knowledge graph.

Set everything: true to delete all user data.
Set dataset or datasetId alone to delete an entire dataset.
Set dataset/datasetId + dataId to delete a single item.
Set dataset/datasetId + memoryOnly: true to clear memory (graph + vector), preserving raw files so the dataset can be re-cognified.
Set dataset/datasetId + dataId + memoryOnly: true to clear memory for a single file only.
Set cleanHistory: true alone to delete only your own search and recall history, or with any target to delete it along with the data.

Request Parameters

dataId (Optional[UUID]): UUID of a single data item to remove. Requires dataset or datasetId to also be set.
dataset (Optional[str]): Name of the dataset to delete or clear.
datasetId (Optional[UUID]): UUID of the dataset, alternative to dataset.
everything (bool): When true, permanently deletes ALL datasets and data the user owns (default: false).
memoryOnly (bool): When true, delete only memory (graph + vector embeddings), preserving raw files and data records (default: false).
cleanHistory (bool): When true, also delete the caller’s own search and recall history before the data is deleted (default: false). Forgetting a dataset or everything purges that data’s history regardless.

Provide either dataset or datasetId, not both. Field names are shown camelCased in the schema; snake_case aliases (data_id, dataset_id, memory_only, clean_history) are also accepted.

Response

A snake_case JSON object whose keys depend on the target (this route answers a union of shapes, so it is not bound to one response model). For a dataId delete it is the deletion receipt — the fields of DeleteDataReceiptDTO in snake_case (status, dataset_id, data_id, data_record_found, deleted_nodes, deleted_edges, data_remaining, dataset_deleted); status is "success", or "not_found" when nothing existed for the id. The two dataset-data DELETE routes return the same receipt camelCased.

Error Codes

404 Not Found: Dataset name does not exist or is not accessible
403 Forbidden: Caller lacks delete permission on the dataset
422 Unprocessable Entity: Invalid parameter combination (e.g. both dataset and datasetId, dataId without a dataset, or memoryOnly without a dataset)
500 Internal Server Error: Error during deletion

POST

api

forget

Forget data: one item, a dataset, its memory only, or everything

curl --request POST \
  --url https://{tenant}.aws.cognee.ai/api/v1/forget \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "dataset": "main_dataset",
  "memoryOnly": true
}
'

import requests

url = "https://{tenant}.aws.cognee.ai/api/v1/forget"

payload = {
    "dataset": "main_dataset",
    "memoryOnly": True
}
headers = {
    "Authorization": "Bearer <token>",
    "Content-Type": "application/json"
}

response = requests.post(url, json=payload, headers=headers)

print(response.text)

{}

{
  "error": "<string>",
  "detail": "<string>"
}

{
  "error": "<string>",
  "detail": "<string>"
}

{
  "error": "<string>",
  "detail": "<string>"
}

{
  "error": "<string>",
  "detail": "<string>"
}

Authorizations

Authorization

string

header

required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Body

application/json

dataId

string<uuid> | null

UUID of a single data item to remove. Requires dataset or datasetId to also be set.

Example:

""

dataset

string | null

Dataset name to delete (or clear with memoryOnly). Provide either dataset or datasetId, not both.

Example:

"default_dataset"

datasetId

string<uuid> | null

Dataset UUID, alternative to dataset. Provide either dataset or datasetId, not both.

Example:

""

everything

boolean

default:false

DANGER: when true, permanently deletes ALL datasets and data the user owns (relational records, graph, vector embeddings, session cache). Ignores dataId/dataset/datasetId.

memoryOnly

boolean

default:false

When True with a dataset, delete only memory (graph nodes/edges and vector embeddings) and reset pipeline status — raw files and data records are preserved. This allows re-cognifying the dataset from scratch.

cleanHistory

boolean

default:false

When true, also delete the caller's own search and recall history (the rows GET /v1/search and GET /v1/recall list), in every dataset, before the data is deleted. Dataset and everything forgets purge the deleted datasets' history regardless of this flag.

Response

Successful Response

The response is of type Response Forget Endpoint Api V1 Forget Post · object.

⌘I

Forget data: one item, a dataset, its memory only, or everything

curl --request POST \
  --url https://{tenant}.aws.cognee.ai/api/v1/forget \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "dataset": "main_dataset",
  "memoryOnly": true
}
'

import requests

url = "https://{tenant}.aws.cognee.ai/api/v1/forget"

payload = {
    "dataset": "main_dataset",
    "memoryOnly": True
}
headers = {
    "Authorization": "Bearer <token>",
    "Content-Type": "application/json"
}

response = requests.post(url, json=payload, headers=headers)

print(response.text)

{}

{
  "error": "<string>",
  "detail": "<string>"
}

{
  "error": "<string>",
  "detail": "<string>"
}

{
  "error": "<string>",
  "detail": "<string>"
}

{
  "error": "<string>",
  "detail": "<string>"
}