Skip to main content
A connected app is a source of content in another service, such as Google Drive or GitHub. You connect the app once, then import the files or pages that you want the chatbot to learn from. The CLI calls a connected app a source. All commands on this page start with sitegpt knowledge sources. knowledge connections does the same. This page shows the common path for each app. For every option, see External data sources. To connect apps in the dashboard, see Connect apps.

Before you start

  • A SiteGPT account on the Growth plan or above. See Plans and limits.
  • The CLI installed and logged in. See Install and log in. The default access from sitegpt login is enough for the tasks on this page.
  • Your chatbot ID. Run sitegpt chatbots list to find it.
  • A web browser, to approve access in the other app. GitHub does not need this.
  • Enough pages left for the new content. Run sitegpt limits to check.

Connect Google Drive, Dropbox, OneDrive, Box, or SharePoint

These apps show a file picker after you approve access.
1

Create the connection

Set --connector to GOOGLE_DRIVE, DROPBOX, ONEDRIVE, BOX, or SHAREPOINT:
The output shows the source ID and an Authorization URL.
2

Approve access and pick files

Open the authorization URL in a browser. Approve access, then select the files in the picker.
3

Check the selected files

4

Import the files

To import only some files, add --external-id with a file ID from the previous step. You can repeat it.
Check that it works: Run sitegpt knowledge documents list --chatbot YOUR_CHATBOT_ID --source GOOGLE_DRIVE. Use the connector name that you chose. Each file has the status SUCCESS when training is done.

Connect Notion

In Notion, you select the pages during the approval, not in a picker.
1

Create the connection

2

Approve access and select pages

Open the Authorization URL from the output. In Notion, select the pages and databases that the chatbot can read.
3

Import the pages

SiteGPT finds the selected Notion pages when you import. So sources documents can show no documents before the first import. This is normal. Check that it works: Run sitegpt knowledge documents list --chatbot YOUR_CHATBOT_ID --source NOTION. Each page has the status SUCCESS when training is done. If no pages show, run sources authorize again and select pages. See Add more files to a connection.

Connect Confluence

1

Create the connection

Pass your Confluence site domain:
2

Approve access

Open the Authorization URL from the output and approve access.
3

Find the space and page IDs

List the spaces, then the pages in one space:
4

Import the pages

Pass one --page for each page:
Check that it works: Run sitegpt knowledge documents list --chatbot YOUR_CHATBOT_ID --source CONFLUENCE. Each page has the status SUCCESS when training is done.

Connect GitHub

GitHub uses a personal access token instead of a browser approval.
1

Create the connection

2

Find the repository

To see the files in one repository, use sources github files with --owner, --repo, and --branch.
3

Import the files

  • --repo is required.
  • --pattern limits the import to matching files. You can repeat it.
  • --sync sets auto-sync to DAILY, WEEKLY, or MONTHLY. GitHub is the only connected app that SiteGPT auto-syncs.
Check that it works: Run sitegpt knowledge documents list --chatbot YOUR_CHATBOT_ID --source GITHUB. Each file has the status SUCCESS when training is done.

Add more files to a connection

To select more files, approve access again, or reconnect after access expired, authorize the connection that you have:
Open the Authorization URL from the output. Then import the new files. reauthorize, select-files, and picker do the same as authorize. Use the connection that you have. A new connection for the same app is a separate connection. Check that it works: sitegpt knowledge sources documents --chatbot YOUR_CHATBOT_ID YOUR_SOURCE_ID lists the new files.

Refresh content from a connected app

SiteGPT auto-syncs only GitHub. For other apps, resync the documents when the files change:
If you pass --sync to sources ingest for another app, SiteGPT sets it to NEVER and the response includes a warning. Check that it works: The CLI prints Documents queued for resync: with a number.

Rename a connection or change the GitHub token

Check that it works: Run sitegpt knowledge sources get --chatbot YOUR_CHATBOT_ID YOUR_SOURCE_ID. The output shows the new name.

Disconnect an app

1

Revoke the connection

sources delete does the same. The chatbot keeps the content that it already imported.
2

Delete the imported content (optional)

To also remove the content, delete the documents from that app:
This deletes the documents from every connection to that app on this chatbot. To delete only some documents, pass their IDs instead. See Delete content.
Check that it works: The revoke command prints the connection details with Status: REVOKED.