Secret & PII Scanner for Confluence and Jira

Privacy and data handling

In short, the app reads page, blog post and issue text inside Atlassian's cloud, keeps only masked findings in Forge storage, and sends nothing anywhere else.

What the app reads

What the app stores (Forge KVS, hosted by Atlassian)

The app stores these items for each page, blog post or issue with findings. Clean content leaves no record.

The app also stores settings, the state of the last scan job, and one random 256-bit secret per installation (used only to compute finding ids, see below).

The full secret or personal value is never stored. Finding ids are SHA-256 hashes of product, content id, field, detector and a keyed fingerprint of the value (HMAC-SHA256 with the installation's random secret). The fingerprint lets the app tell two different secrets apart without storing either; without the installation's secret it cannot be matched to a value.

The app stores no Atlassian account IDs and no user names.

Retention

Who can see findings

Only Confluence administrators (on the Confluence admin page) and users with the Administer Jira permission (on the Jira admin page). The app checks this as the calling user on every request and refuses if the check fails. It also checks, as that admin, which pages and issues they can open; findings in content they cannot open are left out of the list, totals, counts and CSV export.

Egress

None. The manifest has no permissions.external, no remotes and no fetch targets. All calls go to the Confluence and Jira REST APIs of the same site through Forge's requestConfluence / requestJira.