Vulnerability · Published Mar 2, 2026 · Assigned by certcc
CVE-2026-2256Command injection vulnerability in ModelScope's ms-agent
Medium 6.5
In plain English
A medium-severity command injection in ms-agent from ModelScope. Proof-of-concept exploit code is public. The chance it is exploited in the next 30 days is estimated at 1.7% (higher than 77% of all known vulnerabilities). Ready-made attack tooling exists: public proof-of-concept code.
Severity (CVSS 3.1)6.5Medium
Exploit odds (30 days)1.7%top 22.7% of all CVEs
Exploited in the wildNo reportsCISA SSVC: poc
Attack tooling1Public PoC
Official 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.
Affected packages & fixed versions
| Ecosystem | Package | Fixed in | Advisory |
|---|---|---|---|
| PyPI | ms-agent | no fix listed | GHSA-4gc2-344q-r2rw |
| PyPI | ms-agent | no fix listed | PYSEC-2026-2668 |
Affected products (as reported)
| Vendor | Product | Versions |
|---|---|---|
| ModelScope | ms-agent | 0 – through v1.6.0rc1 |
Weakness: CWE-77 · Vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:N
References
- https://github.com/modelscope/ms-agent
- https://www.hiddenlayer.com/research/indirect-prompt-injection-of-claude-computer-use
- https://medium.com/@itamar.yochpaz/cve-2026-2256-from-ai-prompt-to-full-system-compromise-a4114c718326
- https://github.com/Itamar-Yochpaz/CVE-2026-2256-PoC
- https://www.kb.cert.org/vuls/id/431821