CVE-2024-1593: mlflow vulnerable to Path Traversal
MLflow 2.9.2 and earlier mishandle URL parameters. An attacker can smuggle path traversal sequences into the params part of a URL with the ';' character and reach files outside the intended directory. The advisory lists no patched version.
Published Apr 16, 2024 · Analysis Nov 2, 2026
What an attacker can do
A URL has a params section that follows a ';' in the final path segment. The advisory says MLflow's validation does not account for this section, so traversal text placed after a ';' is not seen by the checks but still takes effect when the path is used. It is a variant of earlier MLflow traversal reports, with ';' in the role that '#' and encoded characters play in the others.
The CVSS vector (7.5) shows network access, no privileges and no interaction, with confidentiality loss only. The advisory states that success could lead to information disclosure or server compromise. In practice the attacker reads files the tracking server process can read, such as credentials, keys and configuration, from an exposed server.
Only the tracking server's HTTP surface matters. Local use of the library without a server is not part of the risk.
Who is affected
The pip package mlflow at 2.9.2 and earlier. The sources do not state which later release closes this specific issue. Check the installed version:
pip show mlflowmlflow
- Ecosystem
- PyPI
- Affected
- <= 2.9.2
- Fixed in
- No patched version yet
- Check
pip show mlflow
How to fix it
The advisory lists no patched version and gives no workaround. Review MLflow's release notes before you rely on a newer release. The related traversal fixes shipped in 2.10.0 and 2.12.1, but the sources do not tie this CVE to either.
Until you have confirmed a fix, keep the tracking server off untrusted networks, put an authenticating reverse proxy in front, and run the process with the least file access that works.
How to tell if you were hit
Search proxy and server logs for request paths or parameters that contain ';' followed by '..' or an encoded form, such as %2e%2e, on artifact and model routes. Look for artifact downloads that return files outside your artifact root. Rotate credentials that were readable by the server process if you find such requests.
Original advisory
From the GitHub Advisory Database (CC BY 4.0).
A path traversal vulnerability exists in the mlflow/mlflow repository due to improper handling of URL parameters. By smuggling path traversal sequences using the ';' character in URLs, attackers can manipulate the 'params' portion of the URL to gain unauthorized access to files or directories. This vulnerability allows for arbitrary data smuggling into the 'params' part of the URL, enabling attacks similar to those described in previous reports but utilizing the ';' character for parameter smuggling. Successful exploitation could lead to unauthorized information disclosure or server compromise.
Questions
- What is parameter smuggling in CVE-2024-1593?
The attacker places traversal sequences in the URL params section after a ';'. The advisory says this part escapes validation.
- Is a patch available?
The advisory lists none. Treat servers on 2.9.2 and earlier as exposed.
- Does it need authentication?
No. The CVSS vector lists no privileges and no user interaction.
References
How this page is made: the data panel comes from the GitHub Advisory Database, CISA’s KEV catalog, the NVD and FIRST EPSS, and updates twice a day. The analysis is written by an AI model from the sources listed above and checked by automated rules before it is published. Check versions against the vendor's advisory before you act. Methodology
More MLflow vulnerabilities
All MLflow entriesCVE-2026-116690
MLflow vulnerable to arbitrary file read on unauthenticated servers via prompt model version source
high7.5Oct 27, 2026CVE-2026-116691
MLflow CrewAI autologging records LLM provider API keys in logged trace spans
medium6.5Oct 27, 2026CVE-2026-116692
MLflow vulnerable to remote code execution on unauthenticated servers via third-party scorer deserialization importing caller-placed artifact modules
critical9.8Oct 27, 2026CVE-2026-115886
MLflow: SSRF on unauthenticated MLflow servers via third-party scorer kwargs in the scorer invoke, register, and GenAI evaluate endpoints
medium6.5Oct 23, 2026CVE-2026-53995
MLflow FileStore trace IDs allow path traversal and arbitrary file writes
high8.2Oct 20, 2026GHSA-gqvg-gmmx-x4hm
MLFLOW_ALLOW_PICKLE_DESERIALIZATION=False safety control bypassed by mlflow.statsmodels flavor — RCE via crafted model artifact
high8.8Sep 1, 2026
