Ask who can help.
A sentence in, up to three colleagues out, each with one line of evidence from work they shipped. If nobody here has done it, Keegi says so.
LoadingAsk who can help.
A sentence in, up to three colleagues out, each with one line of evidence from work they shipped. If nobody here has done it, Keegi says so.
LoadingA sentence in, up to three colleagues out, each with one line of evidence from work they shipped. If nobody here has done it, Keegi says so.
Jira is where most workspaces get their evidence. Keegi reads issues from the projects you name and builds each person's card from what they closed.
It reads. It never writes. Every scope the app asks Atlassian for starts with read:, and you can see the whole list on the consent screen before you approve it. A missing scope is a stronger promise than a sentence in a document, which is why that list is short enough to check.
The same approval covers Confluence, if your site has it. Details below.
One thing has to be true, and it is about who you are rather than about setup: you need to be a Jira administrator of the site, or you need one sitting next to you. Atlassian shows a consent screen and only an administrator can approve it for the whole site.
There is nothing to register, no client id to find, and no secret to copy. That is our side and it is already done.
A Jira row appearing in the list is how you know it worked. If you cancel on Atlassian, the page says nothing was connected, and nothing was.
Connecting does not read anything. Keegi lists every project it can see, all of them off, and reads a project only where you turn it on and press Save. Until you do, the Jira row says Keegi reads nothing from this yet.
Above a dozen projects the list gets a filter, and it shows sixty at a time. Type a name to reach the rest. Ticking rows changes nothing on its own, so you can work through a long list and commit once.
New projects appear after Check for new projects. Keegi discovers once when you connect, and a project created next month looks the same to this page as one you are ignoring on purpose.
Turning a project off stops it being quoted. What was already read stays where it is.
Confluence is not a separate connection. Jira and Confluence sit on one Atlassian account, so the consent screen you approve asks for both, and a Confluence row appears beside the Jira one when it does.
Keegi reads pages, attributed to whoever wrote them. Not whoever last edited them. Fixing a broken link on somebody else's runbook is not a runbook you wrote, and a card built the other way would say it was.
Spaces work like projects. Every space starts off. Turn on the ones Keegi may read and press Save. Turning one off deletes what was read from it.
If you connected Jira before September 2026 there is no Confluence row, and pressing Connect Jira again is what adds one. The approval you gave then did not mention Confluence, so Keegi does not have it.
Press Read Jira now. A first read of a large site takes a few minutes, and if it runs out of time what it already read is kept, so pressing again carries on rather than starting over.
Afterwards the row says when it was last read. While a read is partway through it also says how many of your projects are done.
Keegi refreshes its own access when a read or a question needs it. You do not paste anything and there is nothing to renew on a calendar.
A workspace left alone for about 90 days, or connected more than about a year ago, needs Connect Jira pressed again. Atlassian expires the grant, not us.
Disconnect removes the credential. Jira issues that were already read stay until the source is excluded or read again. Confluence pages do not: they are deleted, because a half-read wiki with a saved position is worse than none.