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

# Granola

> Read meeting notes, AI summaries, transcripts, folders, and webhook configuration from Granola

The Granola source reads meeting notes, summaries, transcripts, and folders
that your API key can access. It uses an
[HTTP manifest](/pages/connectors/building-a-connector/http-manifests) and is
**alpha**. It has been checked against Granola's API docs and tested with mock
responses, but has not yet been tested against a live account.

## Configuration

| Field     | Scope      | Default | Description                                      |
| --------- | ---------- | ------- | ------------------------------------------------ |
| `api_key` | Connection |         | Required. Secret. Personal or workspace API key. |

Create a key in **Settings → Connectors → API keys** in Granola. API access
requires a Business or Enterprise plan. On Enterprise, admins choose which
access scopes members can use.

A personal key can read notes covered by the scopes you chose when creating
it. A workspace key can read public workspace notes and notes in spaces with
API access enabled. See [Granola's API guide](https://docs.granola.ai/introduction)
for setup details.

## Resources

| Resource            | Endpoint                             | Parent  |
| ------------------- | ------------------------------------ | ------- |
| `notes`             | `GET /v1/notes`                      |         |
| `note_details`      | `GET /v1/notes/{note_id}`            | `notes` |
| `transcripts`       | `GET /v1/notes/{note_id}/transcript` | `notes` |
| `folders`           | `GET /v1/folders`                    |         |
| `webhook_endpoints` | `GET /v1/webhook-endpoints`          |         |

Notes, note details, transcripts, and folders are selected by default. You can
also select webhook configuration. Filament only writes the resources you
select, though it still lists notes to fetch their details or transcripts.

`notes` contains IDs, titles, owners, and timestamps. `note_details` adds
summaries in text and Markdown, private notes, calendar details, attendees,
and folder membership. Private notes are only returned when the key belongs
to the note's creator. Otherwise, those fields are null. Nested objects and
arrays are stored as JSON.

`transcripts` returns one row per transcript item, with the note ID, speaker,
text, and start and end times. The speaker field keeps any names, attribution,
and anonymous speaker labels Granola provides. Transcripts are fetched in
pages so long meetings can be read in full.

`folders` includes `parent_folder_id` so you can follow the folder hierarchy.
`webhook_endpoints` lists webhook URLs, subscribed events, scopes, folder
filters, and whether each endpoint is enabled. It only reads this configuration.
It does not create webhooks or receive their events.

The [Enterprise Audit API](https://docs.granola.ai/audit) uses a separate
audit-only key and is not included in this connector.

## Modes

All resources support full reads. `notes` also supports incremental reads
using `updated_at`. Each run sends the saved timestamp as `updated_after`,
with a five-minute lookback by default. This re-reads recently updated notes
to help catch changes around the last run. You can adjust the lookback, but
the cursor field is fixed. The checkpoint key is `notes_updated_at` and uses
the `time` comparator.

Details, transcripts, folders, and webhook configuration are read in full.
Use **full replace** for transcripts. Granola does not give transcript items
stable IDs, and speaker names and timestamps are not guaranteed to be unique.

Full replacement also removes deleted notes and refreshes folder membership.
Incremental note reads cannot detect deletions. They can also miss newly
shared notes if those notes were last updated before the lookback window.

## Behavior

* **Auth**: every request uses `api_key` as a bearer token and goes to
  `https://public-api.granola.ai`.
* **Pagination**: notes and folders request 30 items per page. Transcripts
  request 100. Filament passes the returned `cursor` to the next request and
  stops when `hasMore` is false. Details and webhook configuration do not use
  pagination.
* **Rate limiting**: five requests per second, shared across resources in the
  connection. Granola applies its limit per user or workspace, so other clients
  using that quota can cause throttling.
* **Availability**: notes only appear in the API once they have a generated
  summary and transcript. Notes still being processed are not returned.
* The `raw` column keeps any fields that are not mapped to their own columns.