Security overview
Where your data lives, who can see it, how long it is kept, and how we build and run the apps.
This page covers PageCycle and PageCheck: the short version of how each protects your data. PageCycle is built for teams that answer to auditors; PageCheck has its own section below. Our internal policies are available on request through the support portal. To report a vulnerability, see the vulnerability disclosure policy.
PageCycle keeps your data in your Atlassian site
- Runs on Atlassian. PageCycle runs entirely on Atlassian's Forge platform and carries the Runs on Atlassian badge: no servers of ours and no outside services. PageCycle sends nothing to BrainGrid or anyone else.
- Where it lives. Workflows, page statuses and history are stored in Forge SQL inside your Atlassian site, under your site's data residency. Each page's status is Confluence's own status.
- Encryption. Data is encrypted at rest and in transit by Atlassian's platform.
- No sub-processors. No company other than Atlassian ever handles your data.
- We can't see it. BrainGrid AI staff have no access to your site or its data.
People see only what Confluence lets them see
- PageCycle has no logins of its own; Atlassian signs everyone in.
- Every page view, list, count and export shows only pages the person can view in Confluence, and a person's own move needs their edit permission. A page someone can't view never appears for them, not even as a number. The one exception is a space admin's preview of a settings change, whose page counts include every page in the space, as the settings screen's do. A page the admin can't view is never named; a page they can view but not edit is named so they can ask someone who can edit it.
- Only space admins can change a space's workflow, automations and settings.
- On the page, the board, space settings, in Rovo, in AI tools and through the REST API, PageCycle runs with your own Confluence permissions. Confluence checks them itself, and a status you set shows under your name in the page's history. A space admin's space-wide changes (renaming a status, moving pages out of one) run with that admin's permissions too, and leave a page they can't edit as it is.
- Automations, and putting back a status a workflow doesn't allow, run as PageCycle and show as PageCycle in the page's history. When a space admin turns PageCycle on, it reads every page's status to record where each one starts, without changing any page. PageCycle's own History names the person who moved the page, or the automation that did.
- PageCycle's REST API is off until a site or org admin turns on App REST APIs for PageCycle in Connected apps, and turning it off stops every call at once. A script signs in as a person, with an OAuth 2.0 integration that person accepts, and gets answers only for them, with their permissions. Atlassian's gateway names the person; PageCycle never takes it from the request.
- A change to a space's settings through Rovo, an AI tool or the REST API always shows a preview first and applies only when the same space admin confirms it. Rovo asks you to approve each change on its review card. In other AI tools, approve each tool call yourself: an agent that runs tools without asking could confirm a change you never saw.
Personal data
PageCycle stores Atlassian account ids (who moved a page, whose edit caused an automatic move, who created an automation, and which space admin asked for a change through Rovo, an AI tool or the REST API) and the comments people write when they move a page. Nothing else. It reports stored account ids to Atlassian's personal data reporting every week and replaces closed accounts with "Former user". Details: the privacy policy.
Retention and deletion
Space admins choose how long history is kept: 30 days, 90 days, 1 year or 3 years. Older entries are deleted every week. Records of changes asked for through Rovo, an AI tool or the REST API are deleted after 30 days. When you uninstall, Atlassian deletes PageCycle's storage; your page statuses stay, because they are Confluence's own.
How we build and run it
- Every change goes through review and an automated release gate: permission tests, dependency vulnerability audit, and Atlassian's Runs on Atlassian eligibility check.
- Dependencies are scanned on every release and between releases; a software bill of materials (SBOM) is produced for each version and available on request.
- We follow Atlassian's security bug fix deadlines: critical issues fixed within 10 days, high within 4 weeks (the vulnerability disclosure policy has every severity).
- If a security incident ever affects you, we tell Atlassian within 24 hours and, where possible, you within 72 hours, following our incident response plan.
PageCheck
PageCheck is coming soon to the Atlassian Marketplace. This is how it handles your data.
- It runs entirely inside Atlassian on its own model. PageCheck checks the published version of each page inside the app, on Atlassian's Forge platform: page text never leaves Atlassian, and the app calls no outside service. Results and passages are stored in Forge SQL inside your Atlassian site, under your site's data residency.
- Who sees results. Only people who can view a page, checked with Confluence on every request. Nothing is checked until a Confluence admin chooses the spaces PageCheck runs in.
- Personal data. PageCheck stores the Atlassian account id of the person each check started for (who asked for it, who published the edit, or who opened the page when it had no check yet), so every check names who started it. It reports those ids to Atlassian's personal data reporting API every week and removes the id of an account Atlassian reports closed. Page text can contain personal data, such as names written on a page; PageCheck keeps the checked text of the newest two checked versions of each page, only to show the AI check. Details: the privacy policy.
- No sub-processors beyond Atlassian, and BrainGrid AI staff have no access to your site or its data.
Compliance
PageCycle and PageCheck inherit the security and compliance of Atlassian's cloud platform, where your data lives. BrainGrid AI does not yet hold its own SOC 2 or ISO 27001 certification; our security questionnaire answers and internal policies are available on request.
Report a vulnerability
Email security@braingrid.ai. We acknowledge reports within 2 business days. Our vulnerability disclosure policy covers what is in scope, how to test and how fast we fix issues.
Questions
Ask in the support portal.