CVE-2026-121864: Nanobot Sandbox Escape via Caller-Selected Shell Executable
In nanobot, when sandboxed exec is enabled, the shell executable used to run commands can be selected by the caller and is launched before the sandbox is applied. An attacker who can induce the agent to create an executable in the workspace and select it as the shell can cause that executable to run outside the sandbox (e.g. Linux bubblewrap), with the operating-system permissions of the nanobot process.
This issue affects nanobot (nanobot-ai) from 0.2.1 before 0.4.1.
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade
nanobot-aito a version that resolves this vulnerability.Fixed in 0.4.1
Event History
Frequently Asked Questions
Which deployments are affected?
The issue affects nanobot-ai versions from 0.2.1 up to, but not including, 0.4.1 when sandboxed exec is enabled.
What does an attacker need to exploit this issue?
An attacker must be able to induce the agent to create an executable in the workspace and select that executable as the shell used to run commands. The malicious executable then runs before the sandbox is applied.
What access does successful exploitation provide?
The attacker-controlled executable runs outside the sandbox with the operating-system permissions of the nanobot process. This can expose confidentiality, integrity, and availability beyond the intended sandbox boundary.
What should be prioritized for remediation?
Upgrade nanobot-ai to version 0.4.1 or later. Until an upgrade is possible, avoid allowing untrusted parties to cause executable creation in the workspace or to select the shell executable for sandboxed command execution.