Find secrets and personal data in Confluence, and work through them.

Click Scan all pages. The app lists every space, then reads the current version of every page and blog post in each space, with its title and text, and checks them against its rules. A progress line shows the step: Listing spaces, Scanning pages and blog posts (with the space it is on and how many pages it has scanned), and Removing findings from pages that are gone. You can leave the page; the scan carries on.

• Stop scan asks first. Findings saved so far are kept, and the next full scan starts again from the first space.
• If a scan makes no progress for 20 minutes, the page says The scan has stopped moving and offers Start again. Findings saved so far are kept.
• The line next to the button shows the last full scan: when it finished, whether it was started from the page (and by whom) or was the weekly scan, how many pages it scanned and how many had findings.
• Weekly scan. A full scan also starts once a week by itself. It is skipped if a scan is already running.
• On every edit. When a page or blog post is created, edited or restored from the trash, or a page is moved, the app scans that one page again.
• At the end of a full scan, findings on pages the scan did not see (trashed, deleted, archived, or no longer visible to the app) are removed.
• If the app can't read a space, it skips it, says so in a warning after the scan, and keeps that space's earlier findings.

The line above the table counts open findings and the pages they are on, and how many are resolved or marked as false positives. On a large site the counts cover the first 2,000 pages with findings, and the line says so.

• Filters. Space (each with its number of open findings), Type (All, All secrets, All personal data, or one type with its open count) and Status (Open, which is the default, Resolved, False positive or All).
• Space and Page. The page title links to the page in Confluence. Blog posts are marked Blog post. A value found in a title is masked in the title too.
• Type and Value. The value shows its first and last 2 characters, for example ab****yz. Values shorter than 10 characters are fully masked, and so are SSNs.
• Line. The line around the value, with every value on that line masked, then Line 12 or In the title, and 3 times on this page when the same value appears more than once. A note says when a page was so long that only its first part was scanned, or holds more findings than are listed.
• Last edited. Who last edited the page, and when.
• Status. Open, Resolved or False positive, with who marked it and when.

When there are more rows, Load more fetches the next set. How it works at the top of the page gives a short summary of what the app does and doesn't scan.

The app scans every page as itself, so before a row is shown it asks Confluence, as you, whether you can open that page or blog post. If you can't, the row shows only the type of finding, the space, A page you cannot open (id …) and the dates. There is no title, value, line, editor or status, and you can't mark the finding or scan the page again. These findings are still counted at the top, and the page says when some rows are hidden.

If Confluence doesn't answer the check (for example when it is busy), the row is hidden in the same way and the page asks you to refresh and try again. The export follows the same rules.

• Comments, attachments, older versions, drafts and archived pages. Only the current version of each page and blog post is scanned. Whiteboards, databases and text inside images or diagrams are not scanned either.
• Pages and spaces the app can't read. A page restricted so the app can't read it is not scanned. A space the app can't read is skipped, with a warning.
• Very long pages. Only the first 1,000,000 characters are scanned. A page keeps at most 200 findings. The table says when either happens.
• Deleted and trashed pages. Their findings go at the end of the next full scan, or straight away with Rescan page if Confluence still lets you open the page.
• Values the rules don't know. The rules are patterns and checks, not a guarantee. Some formats need a nearby word, as described above, to keep false positives down.