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

# Export and your data

> What export_memory returns (format renor-export/v1: the fact ledger, daily entries, notes and retired items), how paging works, how to import Markdown notes, and how to remove memories.

Your memory is plain Markdown. You can read it without Renor.

## What `export_memory` returns

`export_memory` returns everything a workspace holds, in one documented format (`renor-export/v1`). The reply is two parts:

1. a JSON **manifest**: the ledger's path, the format version, a count of each kind of item (`ledgerEvents`, `dailyEntries`, `notes`, `retired`), the sections this reply carries and, when it stops early, where to continue (`nextSection`, and `nextItem` when a section is cut);
2. a **Markdown document** with the items themselves, in four sections in a fixed order:

   * **ledger** — the fact ledger exactly as stored. Each event has its id and time, its kind (`store`, `supersede`, `retire` or `restore`), its subject and predicate, then its `object` for a fact write or its `target` for a retire or restore (the fact event it acts on). It also has the sentence you said when you gave one, its source, and its scope when that is not global. A `supersede` names the event it closed in `closed_event`, so superseded values stay in the ledger.
   * **daily-entries** — every daily-log entry: what `remember` saves without a subject, predicate and object, and what `remember_note` appends. Each has its id, time, source, scope and text. Identical entries (same time, source and text) appear once.
   * **notes** — every other live Markdown note (imports under `Imported/<label>`, hand-written notes), each in full with its title, modification time and frontmatter metadata.
   * **retired** — everything retired and not restored, newest first: notes, entries and facts, each with when it was retired.

Nothing is summarised: every item is included in full, with the metadata listed above. A workspace with no memories yet returns an empty ledger.

```text theme={null}
## MEM_MUYN8CMZPSI142 · 2026-10-07T21:52:30.967Z
- kind: supersede
- subject: Fjord
- predicate: database
- object: SQLite
- closed_event: MEM_MUYN8C3O7FIQH5
- statement: "We moved Fjord to SQLite because it needs no server."
- source: node
```

## A large workspace is paged, never cut

One reply is bounded at 600,000 characters by default. Without arguments, the reply carries whole sections in order while they fit, and `nextSection` names the first one that does not fit whole. Call again with `section` set to `nextSection` and `offset` set to `nextItem` (0 when the manifest has none), and repeat while `nextSection` is set. When `nextSection` is null, nothing is left. Follow `nextSection`, not `truncated`: when you ask for one section, `truncated` only says whether that section was cut short.

Nothing is dropped, and each call moves forward: it carries at least one item or moves on to the next section, so following the manifest always ends. An item is never split, so one larger than the bound is returned alone, and that reply is longer than the bound. A ledger too large for one reply is paged by line. A continuation's document opens with a heading line such as `# renor-export/v1 · ledger (…, continuing from item N)`; it is not an item, so leave it out when you join the pages.

## What it does not include

<Warning>
  The search index and the embedding vectors are not exported. They are derived from the Markdown and are rebuilt from it, so nothing you can read is lost.
</Warning>

Exporting is not counted against your monthly memory calls. See [Limits and pricing](/limits-and-pricing).

## Import

Import goes the other way, in three steps:

1. `preview_import` shows what would be added: up to 100 Markdown files per call, 64 KiB each, 128 KiB per batch. It writes nothing.
2. `apply_import` adds them under `Imported/<label>` without overwriting anything. Imports create notes, not facts.
3. `undo_import` retires exactly the notes that batch created.

Arguments are on the [Tools](/tools) page.

## Removing memories

* **Retire** one memory with `retire_memory`: it leaves every read and search at once, and `restore_memory` brings it back. Retiring is not erasure: a retired memory is still in your workspace and its export, a retired note keeps counting toward storage, and a retired log entry stops counting when you retire it.
* **Delete permanently** a retired note or log entry on the dashboard's Memories page, one at a time or every retired one shown there. Its text is removed from your workspace, and a retired note's stored copy stops counting toward storage. It is gone from the export and cannot be undone or restored. Only you can do this, from the dashboard; agents cannot. A retired saved fact stays in the history record.
* **Delete your account and memories** by emailing [privacy@renor.cloud](mailto:privacy@renor.cloud) from your account address. It is processed by hand and Renor confirms when it is done. There is no self-service account deletion yet, and a retired fact is erased the same way. Renor's analytics, error-reporting and email providers keep their own logs for their own retention periods; deleting your account does not erase events already sent to them. Paddle, our merchant of record, keeps its own payment records as the law requires; deleting an account cancels the subscription but does not erase those records.

## Where your memory lives

On Renor-controlled infrastructure. The dashboard shows it in your browser when you ask, but it is not stored there, and it is never in a public repository. See [Security](/security).


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