Connect Confluence

A Confluence source keeps a collection in step with the pages of one or more spaces — optionally a single page tree, or only pages carrying certain labels — and, if you want, their attachments. Confluence Cloud and Confluence Data Center are both supported.

Everything Connect a source says about schedules, Sync now, the run log and what happens when a page is edited, removed or deleted applies unchanged. This page covers what is specific to Confluence.

Data Center. The Cloud and Data Center paths make the same calls to the same endpoints; we have verified the Cloud one against a live site. If you connect a Data Center instance and something surprises it, tell us.

Before you start: a credential that can read the spaces

Confluence Cloud: the platform signs in as a user, with an API token. Any user who can read the spaces will do; a dedicated account keeps the audit trail clean.

1. Signed in as that user, open https://id.atlassian.com/manage-profile/security/api-tokens and create a token. Copy it now.
2. Note the user’s email address — it is the username.
3. Your site URL is https://<yoursite>.atlassian.net/wiki — the /wiki matters.

Confluence Data Center: the platform signs in with a personal access token. Create one under the user’s profile → Personal Access Tokens; read permission is enough. The base URL is the Confluence root, for example https://confluence.example.com.

Add the source

Open a collection’s Sources tab, choose Add source, pick Confluence, and fill in:

• Base URL — as above.
• Deployment — Cloud or Data Center. This chooses which credential fields appear.
• Space keys — one per line, for example ENG. Blank means every space the account can read.
• Root page ID — optional. Only this page and its descendants. The id is the number in the page’s URL.
• Labels — optional. Only pages carrying one of these labels.
• Include attachments — on by default. Attachments of the pages in scope become documents too, when their type is supported.
• Under Advanced, Extra CQL — a fragment added to the query the platform builds, for example creator = jsmith. Anything Confluence Query Language accepts.
• Email and API token (Cloud), or Personal access token (Data Center). Tokens are write-only once saved.

Test connection runs the source’s own query with a limit of one — the same query a sync makes — so a wrong token, a base URL missing /wiki, a space the account cannot read and a malformed CQL fragment each fail here with a message that says which.

What becomes a document

Every page the query matches becomes a document, ingested as the rendered page a reader sees, with its title, and cited by its own Confluence URL. Each supported attachment of those pages becomes a document of its own, cited by the page’s attachments view. Blog posts, comments and page restrictions are not read.

How changes are detected

Each run compares a page’s Confluence version number with the one it saw last time; an unchanged page is not fetched. After the first run, only pages modified recently are listed at all (with a day’s margin, so a clock or time-zone difference cannot hide an edit). A page or attachment deleted in Confluence disappears from the listing and its document is paused, then deleted after the grace period.

Test-connection messages

• the email or API token is wrong — regenerate the token; the email must be the token owner’s.
• the personal access token is invalid or expired — create a new one.
• not found: check the base URL (Cloud needs /wiki) and the space keys — the URL is the usual culprit.
• the account cannot read this space or page — grant the account read access in the space’s permissions.
• Confluence rejected the query: … — the Extra CQL fragment does not parse; the message is Confluence’s own.
• connected, but the query matches no pages — the credential works; the space keys, root page or labels select nothing.