Ingest knowledge from Confluence
What gets imported
For the import you run:
- Page titles and bodies, converted from Confluence's storage format to plain text.
- Pages from the space you specify (via space key), or across all spaces your account can see if you leave it blank — whichever the import returns first, up to your Max Pages setting.
What's not imported:
- Comments on pages.
- Attachments. Not fetched at all — file names aren't indexed, and binary content isn't touched.
- Page labels.
- Spaces or pages your Atlassian account can't see.
Setup walkthrough
About 5 minutes:
- Open Knowledge Hub > Add Source > Confluence.
- Generate an API token at id.atlassian.com if you don't have one.
- Enter your Confluence URL (e.g.
https://yourcompany.atlassian.net), your Atlassian account email, and the API token. - Optionally set a Space Key (e.g.
ENG) to scope the import to one space. - Set Max Pages (1–500; defaults to 50).
- Click Import Pages.
This same email + API token flow is what Confluence Cloud calls "API tokens for personal use" — it works whether or not your organization also has Confluence Data Center, as long as the URL you enter is reachable from AskVault and answers Confluence's REST API.
Refreshing content
There's no background sync. To pick up edits made in Confluence:
- Run the import again with the same space key.
- Pages are matched by URL: unchanged pages are left alone, and pages whose content changed are re-indexed with the new version.
Space scoping
- Space Key set. Only pages from that space are considered, up to Max Pages.
- Space Key blank. Pages across every space your account can see are considered, up to Max Pages — for a large Confluence instance this can pull from unexpected spaces, so scoping to a space key is the safer default.
To index multiple specific spaces, run the import once per space key.
Permission inheritance
Confluence's own permission model applies:
- AskVault reads with the connecting account's permissions — whatever that Atlassian account can see via the API, the import can read.
- A page restricted at the Confluence level isn't returned by the API and so isn't indexed.
Combine with AskVault audience tags on the resulting documents for additional segmentation after import.
Limits
- Pages per import. 500 max, default 50.
- Comments, attachments, labels. Not imported.
- Sync. None — re-run the import manually to refresh.
Common pitfalls
"Confluence authentication failed." Double-check the email matches the Atlassian account the API token belongs to, and that the token hasn't been revoked.
Pages missing from the import. Either they're outside the space you scoped to, past the Max Pages cutoff, or the connecting account doesn't have access to them in Confluence.
Stale answers after editing in Confluence. The import doesn't auto-refresh. Re-run it to pick up the changes.
Self-hosted Confluence unreachable. AskVault must be able to reach the URL you entered over HTTPS.
FAQ
Does this use OAuth?
No — email plus an Atlassian API token, entered once per import. AskVault doesn't store a persistent connection or auto-refresh a token.
Can I index public Confluence spaces without a token?
Not through this integration; it requires a valid email/API-token pair either way. For a public-facing Confluence site, URL crawling is an alternative.
How fresh are bot answers?
As fresh as your last import. There's no scheduled sync — re-run the import when Confluence content changes materially.
Can I limit which pages get pulled in?
Yes, by space key. There's no glob/path filter within a space today — everything in the space (up to Max Pages) is a candidate.
Does this work with Jira too?
Jira is a separate integration with its own import flow. See Jira ingest.