CVE-2026-2256 - Vulnerability Analysis

Overview

ModelScope ms-agent <= v1.6.0rc1 contains a command injection caused by crafted prompt-derived input, letting attackers execute arbitrary operating system commands remotely, exploit requires crafted input.

Severity & Score

Impact

Attackers can execute arbitrary operating system commands, potentially leading to full system compromise.

Mitigation

Update to the latest version beyond v1.6.0rc1.

References

• 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
• https://github.com/Itamar-Yochpaz/CVE-2026-2256-PoC
• https://github.com/modelscope/ms-agent

Social Media Activity(1 post)

LLMs

@LLMs

Mar 13, 2026

Execwall – firewall to stop ModelScope CVE-2026-2256 (AI agent command injectn) CVE-2026-2256 just dropped - a prompt injection in ModelScope's ms-agent allows arbitrary OS command execution.... https://news.ycombinator.com/item?id=47371292 | https://awakari.com/sub-details.html?id=LLMs | https://awakari.com/pub-msg.html?id=LM1LLcb0flHp78GpzsxCqy70ASW&interestId=LLMs

View original post

GitHub Repositories(1 repo)

• https://github.com/Itamar-Yochpaz/CVE-2026-2256-PoC

Related Resources

• Vulnerability Intelligence Dashboard
• Credential Scanner