PT-2025-43576 · Ms-Swift · Ms-Swift

CVE-2025-41419 · Published 2025-07-31 · Updated 2026-07-07

Positive Technologies (AI Summary)

CVSS v4.0

5.1

Medium

Base vector

AV:N/AC:L/AT:P/PR:N/UI:A/VC:H/VI:L/VA:N/SC:N/SI:N/SA:N/E:P

Name of the Vulnerable Software and Affected Versions ms-swift (affected versions not specified)

Description Command injection is possible through the LLM Training interface of the web-ui. An attacker can manipulate the --output dir parameter to execute arbitrary commands on the host system. This occurs because the train() method in swift.ui.llm train.llm train.LLMTrain directly concatenates user-supplied parameters into system commands, and the train local() method in swift.ui.llm train.llm train.LLMTrain executes these commands using os.system().

Recommendations Modify the train() method in swift.ui.llm train.llm train.LLMTrain to stop directly concatenating parameters with commands received from the frontend. Replace the use of os.system() in the train local() method of swift.ui.llm train.llm train.LLMTrain with subprocess.run() using a list of arguments to ensure safe command execution.

Fix

Weakness Enumeration

CWE-117

Related Identifiers

CVE-2025-41419

GHSA-7C78-RM87-5673

PYSEC-2026-1682

Affected Products

Ms-Swift

References · 8

• https://osv.dev/vulnerability/GHSA-7c78-rm87-5673 · Vendor Advisory
• https://osv.dev/vulnerability/PYSEC-2026-1682 · Vendor Advisory
• https://nvd.nist.gov/vuln/detail/CVE-2025-41419 · Security Note
• https://github.com/modelscope/ms-swift · Note
• https://github.com/modelscope/ms-swift/commit/32f09e9b0a44f19d44210e2b5b47c58ab01740e1 · Note
• https://github.com/modelscope/ms-swift/security/advisories/GHSA-7c78-rm87-5673 · Note
• https://github.com/advisories/GHSA-7c78-rm87-5673 · Note
• https://pypi.org/project/ms-swift · Note