Skip to main content
You can load memories into MemoryPlugin from a CSV or JSON file, and take them back out as CSV, JSON, or plain text. Your data is yours: import it, export it, move it, keep your own copy.

Import memories

Import loads memories from a file into one bucket at a time: the bucket you have selected. The import dialog confirms the target (“Importing to …”) so you always know where the memories are going.
Import Memories dialog showing a parsed CSV file with 142 memories found and an Import 142 Memories button

Supported formats

CSV

One memory per row, with a header

JSON

An array of memory objects
The dropzone accepts .csv and .json. The format is auto-detected from the file.

CSV format

Your CSV needs a header row with a text column. That column is the only requirement; a file without it is rejected with “CSV must have a ‘text’ column”.
text
"My preferred programming language is Python"
"I work at Acme Corp as a software engineer"
"I prefer dark mode in all applications"
You can also include a createdAt (or created_at) column to set each memory’s timestamp. Any other columns are ignored.

JSON format

JSON imports expect an array of objects, each with a text field:
[
  { "text": "My preferred programming language is Python" },
  { "text": "I work at Acme Corp as a software engineer" }
]
An optional createdAt field sets the timestamp. Other fields are ignored.

How to import

1

Select the target bucket

In the dashboard, select the bucket you want the memories in (or use General). Import always writes to the selected bucket.
2

Open the import dialog

Click Import in the memory list header. The Import Memories dialog opens.
3

Choose your file

Click to select a .csv or .json file. MemoryPlugin parses it and shows how many memories it found.
4

Import

Click Import Memories. When it finishes, an Import Complete summary shows how many were imported, how many duplicates were skipped, and how many entries were invalid.
Duplicates are skipped automatically. Memories that already exist are not imported again, so re-running the same file will not create duplicates. If the bucket uses Smart Memory, categories are assigned to the new memories automatically.

Bulk add (no file needed)

For a handful of memories you do not want to put in a file, use bulk add instead:
1

Open the Add Memory dialog

Press ⌘J / Ctrl+J, or click New Memory in the sidebar.
2

Turn on bulk mode

Check Bulk mode (one per line).
3

Paste your memories

Put one memory per line, then save. Each line becomes its own memory.

Export memories

Export is opened from the memory list header. Export covers the current bucket; if you have selected specific memories first, the button becomes Export Selected () and only those are exported.
Export Memories dialog with a Copy to Clipboard button and CSV, JSON, and Text download options

Copy to clipboard

The primary export action is Copy to Clipboard. It copies your memories along with AI instructions, ready to paste straight into a chat with any AI tool. Use this when you want another AI to read your memories directly, not when you want a file.

Download as a file

To save a file instead, pick a format under “Or download as:“:
FormatBest for
CSVSpreadsheets and simple backups
JSONProgrammatic use and re-importing later
TextReading, or pasting into anything that takes plain text
A CSV export has the columns text, createdAt, bucketName, and categoryName, so it carries each memory’s text, timestamp, bucket, and Smart Memory category.

Export specific memories

1

Select memories

Use Select Multiple in the memory list header, then check the memories you want.
2

Export the selection

Click Export Selected ().
3

Copy or download

Copy to clipboard with AI instructions, or download as CSV, JSON, or Text.

Use cases

  • Backup. Export a bucket now and then and keep the file somewhere safe.
  • Migration. Export from one account, import into another.
  • Sharing. Hand a colleague an export, or use Copy to Clipboard to paste your memories into their AI chat. (For live shared access, see bucket sharing.)
  • Analysis. Export to CSV and dig through your memories in a spreadsheet.

Tips

Fix formatting and drop obvious junk in the spreadsheet first. It is easier than editing memories one at a time afterward.
There is no per-row bucket column. If you want memories split across buckets, import each set with the right bucket selected, or move them afterward.
A quick export is cheap insurance. It is your data; keep a copy.
Try a few rows first to confirm the format parses before importing thousands.

Next steps

Memory Buckets

Organize your imported memories into buckets

Bulk Operations

Select, move, and delete many memories at once