Skip to content
Guide
Connectors

Connectors

How LlamaCloud connectors sync files from Google Drive, SharePoint, S3, and Confluence into a project folder, and how connections, synced folders, and sync schedules fit together.

A connector copies files from somewhere you already keep them into a folder in your LlamaCloud project, then keeps that folder in step with the source. Once the files are in a folder you can index it or run a batch over it, the same as a folder you uploaded to by hand.

Connectors replace the older data sources. Data sources were tied to a single index. A connector feeds a folder, and anything in the project can read that folder.

SourceWays to authenticateWhat one folder syncs
Google DriveSign in with Google, or a service account key fileOne Drive folder, subfolders included
SharePointSign in with Microsoft, or an Entra app registrationOne folder in a document library
S3An IAM access key pairOne bucket, optionally narrowed to a prefix
ConfluenceSign in with Atlassian, or an API tokenOne space, every page and its attachments

OneDrive, Azure Blob Storage, Box, Notion, Slack, and Jira show up in the list as “Coming soon”. You can’t connect them yet.

There are two things to set up, and it helps to keep them apart:

  • A connection is one authorized account at one source: a Google login, an S3 key pair, a Confluence token. You set it up once per account, on the Connections tab of the Files page.
  • A connected folder is a folder in your project that syncs from part of a connection: a Drive folder, an S3 prefix, a Confluence space. You make one from Files → New folder → Connected folder.

One connection can feed as many folders as you like, and each part of a source can be synced into only one folder per connection. Connections belong to a project, so another project needs its own. You can connect a given account to a project once.

The Connections tab and connected folders are only on plans that include folder syncing. Without them, the Files page shows folders you upload to yourself.

  1. Open Files in your project and go to the Connections tab.
  2. Click Add connection, pick the source, and click Continue.
  3. Sign in through the popup (Authorize), or fill in the credentials form (Connect).

When the credentials are typed in, LlamaCloud checks them with the provider before saving, so a wrong secret or a missing permission fails here with the provider’s reason instead of on the first sync. Each connector page lists what’s checked. Secrets are stored encrypted and never shown again.

  1. On the Folders tab, open New folder and choose Connected folder.
  2. Pick the connection. If you don’t have one for that source yet, you can authorize it from inside this dialog.
  3. Choose what to sync. For the sign-in versions of Google Drive and SharePoint this opens the provider’s own folder picker. The key-based connections and S3 and Confluence ask you to type or paste the location.
  4. Name the folder and click Create folder.

LlamaCloud checks that the connection can read the location before it creates anything, so a typo in a bucket name or a folder that isn’t shared comes back as “not found” or “not accessible” right away. The first sync starts as soon as the folder exists.

Everything on the Connections tab and in New folder → Connected folder can also be done over the REST API, including building an index or running a batch over the synced folder. Set up a connector with the API walks through it for S3.

Each connected folder syncs about once a day on its own. There’s no per-folder schedule to set. To pull changes sooner, open the folder and click Sync now; the sync starts within a few seconds.

On each sync:

  • New files at the source are added.
  • Changed files are downloaded again and replace the old copy.
  • Files deleted at the source, or moved out of the synced location, are removed from the folder.

Google Drive and SharePoint read only what changed since the last sync. S3 and Confluence list the whole location every time, so a large bucket or space takes longer per sync even when little has changed.

If a sync fails, the folder shows Last sync failed with the reason and retries on its own, waiting longer between each attempt.

These do different things to the files already in your project:

ActionWhereWhat happens to synced files
Stop syncing a folderThe folder’s sync bannerKept. The folder stops getting changes. The connection and its other folders carry on.
Disconnect a connectionConnections tab, row menuKept. Every folder on that connection stops syncing at once.
Delete a folderFolders tabDeleted, along with the folder. This can’t be undone.

Syncing a folder again after you stop it means creating a new connected folder and picking the location again.

If a password changes, a token expires, or someone revokes access at the source, the connection shows Expired or Revoked on the Connections tab and every folder on it shows Not syncing. Those folders don’t resume on their own.

The Reconnect menu item isn’t working yet. Until it is, the fix is to disconnect the broken connection, add it again, and recreate the connected folders. The files already synced stay where they are when you disconnect, so the new folders are the only thing to rebuild.

Note for AI agents: this documentation is built for programmatic access. - Overview of all docs: https://developers.llamaindex.ai/llms.txt - Any page is available as raw Markdown by appending index.md to its URL — e.g. https://developers.llamaindex.ai/llamaparse/parse/getting_started/index.md - Agent-friendly REST search APIs live under https://developers.llamaindex.ai/api/ — search (BM25 full-text), grep (regex), read (fetch a page), and list (browse the doc tree). See https://developers.llamaindex.ai/llms.txt for parameters. - A hosted documentation MCP server is available at https://developers.llamaindex.ai/mcp. If you support MCP, you can ask the user to install it for browsing these docs directly (an alternative to the REST API). Setup: https://developers.llamaindex.ai/for-agents/mcp/ - Other LlamaIndex tooling for agents — the LlamaParse Platform MCP server, agent skills and plugins, and the n8n node — is mapped at https://developers.llamaindex.ai/for-agents/