CVE-2023-1177

Deeper analysis

CVE-2023-1177 is a path traversal vulnerability, identified as CWE-22 and CWE-29, that affects the MLflow machine learning platform in the GitHub repository mlflow/mlflow prior to version 2.2.1. The flaw permits directory traversal sequences such as '\..\filename' and carries a CVSS 3.1 score of 9.3 reflecting network attack vector, low complexity, and no required privileges or user interaction.

An unauthenticated remote attacker can exploit the issue to read arbitrary files on the server, resulting in high confidentiality impact and limited integrity impact within the affected MLflow deployment. The vulnerability is exposed through the application's handling of file paths, enabling direct access to sensitive data outside intended directories.

References to the fixing commit 7162a50c654792c21f3e4a160eb1a0e6a34f6e6e and the associated huntr.dev report confirm that the issue was resolved by the MLflow 2.2.1 release; practitioners should upgrade immediately and validate that path-handling logic no longer accepts traversal sequences.

The EPSS score remains elevated at a current value of 0.9331 with a peak of 0.9333, indicating sustained exploitation interest after disclosure for this MLflow component.