CVE-2026-51856 Code injection in n/a
Remote attackers can execute arbitrary Python code by prompting the agent to use the exposed execute_python_code tool without proper isolation.
- Exploited in the wild
- Not known
- Login needed
- Not needed
- Attackable
- Over the internet
- Victim action needed
- Not needed
- Fix
- None listed
What is CVE-2026-51856?
CVE-2026-51856 is a critical code injection vulnerability in n/a, published on 30 Sep 2026. An attacker can exploit it over the internet without logging in, with no action from the victim.
Official description
In agentscope 1.0.18, 1.0.19, and 1.0.19 when the RealtimeAgent session exposes execute_python_code as an available tool, a remote WebSocket user can prompt the agent to call that tool and run Python code in the service environment. In the validated path, RealtimeAgent._acting forwards the model-produced tool call to Toolkit.call_tool_function, which invokes execute_python_code without an additional approval or isolation boundary on that path.
Weakness: CWE-94
Which versions are affected by CVE-2026-51856?
How do I fix CVE-2026-51856?
No fix has been published for CVE-2026-51856 yet. Limit exposure of the affected software until one is available.
Upgrade, and never evaluate templates, expressions or code built from user input. More about Code injection.
Is CVE-2026-51856 being exploited?
It is not confirmed as exploited in the wild, but public exploit code for CVE-2026-51856 exists.
Sources
Also known as GHSA-8v8q-vxfm-pcf6.