Security and data
Runs on Atlassian
Apifolio is built on Atlassian Forge and is eligible for Runs on Atlassian: the app has no external servers and does not send data outside Atlassian. Its code runs on Atlassian's infrastructure, and your data stays under Atlassian's security, compliance and data-residency controls.
How data flows
- A reader opens a page. Apifolio, running in a sandboxed frame inside Confluence, asks Confluence for the spec file as that reader. If they can't see the file, they can't see the docs.
- The file is parsed and rendered in the reader's browser.
- For PDF/Word exports, the Rovo agent and anonymous visitors, Apifolio functions read the same file on Atlassian's infrastructure. Exports and agent answers use the current user's permissions; the anonymous fallback only serves files attached to the page being viewed, up to about 4 MB.
- In the macro editor, Apifolio works as the editor: it lists the page's attachments, reads the page body to find other API-doc apps' macros it can import, searches page titles for "Another page", and saves the files the editor uploads, drops or pastes as page attachments.
- Macros that show the same file on one page can share the downloaded text inside the reader's browser. Each macro first checks the file as the reader, and nothing is stored.
- The Pro tools work in the browser too: version comparison downloads earlier versions of the file and the API catalog uses Confluence search, both as the current user.
Nothing is stored outside Confluence. The macro settings live in the page; the search digest (API title, endpoint paths, operation IDs and summaries) is stored in the macro. The reader's choice of code language is remembered in their own browser.
Permissions (scopes)
| Scope | Why Apifolio needs it |
|---|---|
read:page:confluence |
Read page titles and bodies: the macro settings (agent and export) and, in the macro editor, other apps' API macros to import |
read:blogpost:confluence |
The same for blog posts |
read:attachment:confluence |
List and read the spec files attached to pages, and their versions |
write:attachment:confluence |
Save a spec as an attachment, only when an editor uploads, drops, pastes or imports one |
read:content-details:confluence |
Needed by the Confluence REST API endpoints Apifolio uses to upload an attachment and to run a CQL search |
search:confluence |
Let the Rovo agent (and the API catalog of the Pro edition) find pages with API docs, and the macro editor find pages by title, as the current user |
Safe rendering
Spec files are treated as untrusted input:
- Markdown in descriptions is sanitised (no scripts, frames, forms or styles); external images are shown as links;
- code and examples are rendered as text, never as HTML;
- parsing has size and depth limits, and is protected against hostile YAML (alias bombs) and prototype pollution;
- external
$reffiles and URLs are never fetched.
No credentials, no live calls
Apifolio has no "Try it out" button. It never stores API keys or tokens and never calls the APIs it documents. Code samples use placeholders such as YOUR_TOKEN.
Readers can type their own values, credentials included, into Edit request to fill in the code samples. Those values stay in the memory of the reader's browser tab: they are never saved or sent, and they are gone when the editor closes. What is typed in the credential fields is also masked on screen and left out of Copy for AI.
When an editor imports another app's API macro, credentials stored in it are never copied: the editor only sees their names, as a reminder to rotate them.
Reporting a vulnerability
Write to [email protected] with the details and steps to reproduce. We acknowledge reports within 1 business day and fix confirmed issues within the timelines set by Atlassian's security requirements for Marketplace apps (critical within 10 days, high within 4 weeks, medium within 12 weeks, low within 25 weeks). Please don't disclose the issue publicly before it is fixed.