AgentScope MCP Workspace API OS command injection flaw

An AI agent tool can have commands run on it by remote attackers.

Summary

A flaw in AgentScope's MCP Workspace API lets a remote attacker run operating system commands through an argument. Versions up to 2.0.11 are affected, and exploit code is public.

Affected when

MCP Workspace API in use

What to do

Check that you run a version newer than 2.0.11 and read the vendor's advisory. The fix patch is still awaiting acceptance.

Products and platforms

AgentScope

CVE IDs

CVE-2026-123022 (opens in a new tab)

Sources

Write-up: vuldb.com (opens in a new tab) · vuldb.com (opens in a new tab) · vuldb.com (opens in a new tab) · github.com (opens in a new tab)
CVE record: cve.org (opens in a new tab)

Times

First published 2026-11-05 13:15 UTC

First seen 2026-11-09 16:40 UTC