Atlassian issue trackers hold two kinds of reusable knowledge, and BeforeQuery indexes both:
  • Jira issues + comments — engineering context, workarounds, and decisions recorded on issues.
  • Resolved Jira Service Management (JSM) requests — solved service-desk tickets, indexed so future questions get instant answers.

​
Prerequisites

  • Your Atlassian site URL (e.g. https://acme.atlassian.net).
  • An Atlassian account with access to the projects/service desks you want indexed, plus an API token from id.atlassian.com → Security → API tokens.
  • A BeforeQuery workspace owner role.

​
Data ingested

One request, one document:
  • Request URL on your site, and DESK-123: summary as the title
  • Status and labels
  • The request description, under a ## Request heading
  • Every comment, under ## Replies, each with its author
Only requests matching the filter are fetched — resolved ones by default.

​
Index Jira issues and comments

1

Add the source

In the dashboard, open Knowledge Base → Sources → Add source and choose Jira. Enter the site URL, account email, and API token. Optionally list project keys (e.g. SUP, ENG) to limit indexing; leave empty to index all accessible projects.
2

Sync and verify

Click Add. Issues and their comment threads are normalized into documents. Test in the Playground with a question that was answered in an issue’s comments.

​
Index resolved JSM requests

A service desk is its own source type. It uses Jira’s credential, because a service desk is a Jira project — what differs is the selection and the default: only requests that reached a resolution are indexed.
1

Add the source

Open Knowledge Base → Sources → Add source and choose Jira Service Management. Enter the site URL, account email and API token.
2

Pick the service desks

List service desk projects (e.g. SUP, IT), or leave it empty to index every service desk the credential reaches. Click Test connection, then Add.
Leave resolved requests only on. An open request is a question with no answer under it, and indexing it teaches the assistant to repeat the question back.
Each request becomes one document: the summary, its status and resolution, the request body, and the agent replies under a Replies heading. The replies are the point — the description is the problem and the thread is the solution.
This is separate from the workspace-level helpdesk integration, which powers copilot drafts and live ticket deflection. You do not need one to index a service desk.
Service-desk requests often contain requester PII — enable PII masking before syncing.

​
Keeping content in sync

Both sources re-sync every 10 minutes — there is no schedule to configure, and Sync on the source’s page runs one now if you cannot wait. Only changed content is re-embedded. See Sources Overview.

​
Troubleshooting

  • Missing issues? The connector account lacks project access, or the project key wasn’t included. Adjust and re-sync.
  • Auth errors? API tokens are per-account — make sure the email matches the token owner and the token hasn’t been revoked.
  • Internal engineering context leaking? Mark issue-derived documents restricted or keep the Jira source in an internal-only source group.