- CVSS
- MEDIUM · 6.5v3.1CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:N
- Published
- 2026-03-02
- Weakness
- CWE-77
- Source
- nvd.nist.gov/vuln/detail/CVE-2026-2256
Description
A command injection vulnerability in ModelScope’s ms-agent versions v1.6.0rc1 and earlier exists, allowing an attacker to execute arbitrary operating system commands through crafted prompt-derived input.
References
- https://github.com/Itamar-Yochpaz/CVE-2026-2256-PoC
- https://github.com/modelscope/ms-agent
- https://medium.com/@itamar.yochpaz/cve-2026-2256-from-ai-prompt-to-full-system-compromise-a4114c718326
- https://www.hiddenlayer.com/research/indirect-prompt-injection-of-claude-computer-use
- https://www.kb.cert.org/vuls/id/431821
How GTK Cyber trains on this
AI security training at GTK Cyber covers the LLM and ML-pipeline vulnerability classes that vulnerabilities like CVE-2026-2256 fall into. Our hands-on courses are taught by Charles Givre and other practitioners who break and defend production AI systems.