Data Loss Prevention
Scans Confluence pages and attachments every 10 minutes for exposed login details, personal data, and classified content.
Overview
Confluence is a shared workspace where teams write pages and upload files. Those pages and files can hold login details, personal data, or content marked confidential, and nothing was checking them. The content also had to stay in-house, so sending it to an outside scanning service wasn't an option.
Approach
The company ran its own copy of Confluence, so the scanner could read pages and attachments directly. Every 10 minutes it checks only what was created or changed since the last run. When a match is almost certainly real, the system restricts access to the content and emails the author right away. Everything else waits for a person to review and confirm it.
Architecture
The scanner first pulls the text out of attachments, so a file can be checked the same way as a page. It then looks for known patterns and keywords, and follow-up checks weed out false alarms. The same app lets the team manage the rules and follow results on dashboards. Authors get a self-service page, linked from the email, where they confirm their fix.
Outcome
Over a year it caught thousands of issues across tens of thousands of Confluence spaces and hundreds of thousands of pages, and roughly 15% of its flags were false alarms. Authors who didn't fix flagged content got reminders, and issues that stayed unresolved went to their managers.
Stack
Oracle APEX is a platform for building apps inside an Oracle database, so the scanner and the data it handled stayed in one place. The Oracle tools listed here covered the rest: reading attachments, searching text, connecting to Confluence, sending emails, and running scans on a schedule.