Executive brief
Kolibri, an offline learning platform, contains a vulnerability where certain internal web components do not properly verify the destination of network requests. An attacker can use this to force the Kolibri server to connect to other systems on the internal network or access sensitive cloud configuration data. This could lead to the exposure of internal service information or unauthorized access to private network resources.
Technical details
Multiple API endpoints in Kolibri, including RemoteFacilityUserViewset and RemoteFacilityUserAuthenticatedViewset, fail to validate the 'baseurl' parameter. The root cause is a combination of response reflection (returning remote JSON bodies verbatim) and insufficient restriction on remote targets, as the application followed HTTP 30x redirects by default. An unauthenticated attacker can provide a malicious URL that redirects the server to internal network services or cloud metadata endpoints (like AWS IMDSv1). If the target returns JSON, the response is reflected back to the attacker. The vulnerability is mitigated in version 0.19.4 by implementing response sanitization, mandatory authentication, cross-host redirect blocking, and a peer allowlist.
Affected products
- Learning Equality Kolibri <= 0.19.3
Timeline
- 2026-05-27: advisory: Initial GitHub Advisory published
- 2026-06-11: disclosed: Advisory updated with CVE-2026-48053
- 2026-06-11: patched: Fix released in version 0.19.4
References
- https://api.github.com/users/beraoudabdelkhalek
- https://github.com/beraoudabdelkhalek
- https://api.github.com/users/beraoudabdelkhalek/gists%7B/gist_id%7D
- https://api.github.com/users/beraoudabdelkhalek/repos
- https://avatars.githubusercontent.com/u/89158254?v=4
- https://api.github.com/users/beraoudabdelkhalek/events%7B/privacy%7D