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

# Pagination

> Every list endpoint uses the same cursor-based contract.

All list endpoints return the same envelope:

```json theme={"system"}
{
  "object": "list",
  "data": [ ... ],
  "has_more": true,
  "next_cursor": "eyJkIjoiMjAyNi0wOC0yOSIsImkiOiJhOWIyIn0"
}
```

Request the next page by passing `next_cursor` back as `cursor`. When `has_more` is `false`, `next_cursor` is `null` and you have everything.

<Note>
  `/accounts` is the one list that does not page. It returns every account in one response and accepts neither `limit` nor `cursor`, but keeps the envelope above with `has_more` set to `false`, so code written for the paged endpoints works on it unchanged.
</Note>

```bash theme={"system"}
curl "https://api.thinkout.io/v1/transactions?limit=100&cursor=eyJkIjoiMjAyNi0wOC0yOSIsImkiOiJhOWIyIn0" \
  -H "Authorization: Bearer $THINKOUT_API_KEY"
```

## Parameters

| Parameter | Default | Max   | Notes                                                               |
| --------- | ------- | ----- | ------------------------------------------------------------------- |
| `limit`   | `100`   | `500` | Page size.                                                          |
| `cursor`  |         |       | Opaque token from the previous page. Do not construct or decode it. |

Keep the filters identical between pages. A cursor is only valid for the query that produced it, and reusing it with different filters returns `400 invalid_cursor`.

## Ordering

Each resource has one fixed order. There is no `sort` parameter.

| Resource                                | Order                            |
| --------------------------------------- | -------------------------------- |
| transactions, forecasts                 | `date` descending, then `id`     |
| categories                              | the order ThinkOut displays them |
| accounts, banks, counterparties, labels | `name` ascending, then `id`      |

Categories are the exception. They come back as the product's own category screen reads: inflows before outflows, each root immediately followed by its children, siblings in `position` order. Keep the rows in the order you received them and the tree renders correctly without sorting.

## Why cursors

Bank syncs insert transactions while you page. Offset pagination would skip or repeat rows in that case. The cursor is anchored to the last row you received, so a page never overlaps the previous one.

## Counts

There is no `total_count` in the envelope. To count rows, page to the end. For aggregates, use the reporting features of ThinkOut itself.