An information disclosure vulnerability exists in the lunary-ai/lunary, specifically in the runs/{run_id}/related endpoint. This endpoint does not verify that the user has the necessary access rights to the run(s) they are accessing. As a result, it returns not only the specified run but also all runs that have the run_id listed as their parent run. This issue affects the main branch, commit a761d833. The vulnerability allows unauthorized users to obtain information about non-public runs and their related runs, given the run_id of a public or non-public run.
The product implements access controls via a policy or other feature with the intention to disable or restrict accesses (reads and/or writes) to assets in a system from untrusted agents. However, implemented access controls lack required granularity, which renders the control policy too broad because it allows accesses from unauthorized agents to the security-sensitive assets.
| Name | Vendor | Start Version | End Version |
|---|---|---|---|
| Lunary | Lunary | 1.4.9 (including) | 1.4.9 (including) |
Integrated circuits and hardware engines can expose accesses to assets (device configuration, keys, etc.) to trusted firmware or a software module (commonly set by BIOS/bootloader). This access is typically access-controlled. Upon a power reset, the hardware or system usually starts with default values in registers, and the trusted firmware (Boot firmware) configures the necessary access-control protection. A common weakness that can exist in such protection schemes is that access controls or policies are not granular enough. This condition allows agents beyond trusted agents to access assets and could lead to a loss of functionality or the ability to set up the device securely. This further results in security risks from leaked, sensitive, key material to modification of device configuration.