Executive brief
Lunary is an open-source platform for managing AI application runs and analytics. A vulnerability in the runs API endpoint allowed unauthorized users to retrieve information about runs they should not have access to, including private runs and their related runs. This could expose sensitive AI application data and operational details to unauthorized parties.
Technical details
An information disclosure vulnerability exists in the lunary runs/{run_id}/related endpoint that fails to validate user access rights before returning run data. The endpoint returns not only the requested run but also all child runs with the specified run_id as their parent, regardless of authorization status. An authenticated attacker with knowledge of any run_id (public or private) can enumerate and retrieve related runs they are not authorized to access. The vulnerability affects all versions prior to 1.4.10, which implements proper access control checks on this endpoint.
Affected products
- Lunary AI lunary prior to 1.4.10
Timeline
- 2024-09-13: disclosed
- 2024-09-13: patched: Version 1.4.10 addresses the vulnerability with proper access control checks
- 2025-06-20: other: Advisory withdrawn due to package ecosystem mismatch; vulnerability remains valid for the affected component