Horizon Alert
Summary of the vulnerability and why it matters
This vulnerability in Windmill allows authenticated users to modify critical system files within their sandboxed script executions. This can lead to significant compromise by redirecting network traffic, intercepting sensitive information, and enabling unauthorized access to other users' workspaces.
- Authenticated users can abuse sandbox permissions.
- Enables man-in-the-middle and token theft.
- Leads to workspace takeover.
Attack Path
How an attacker could exploit the issue
An authenticated user with script execution privileges can abuse this vulnerability to tamper with system files inside a sandbox. This allows them to redirect network traffic, intercept sensitive data like JWTs, and impersonate other users to gain administrative access.
- Requires authenticated user access.
- Exploitable via script execution.
- Allows man-in-the-middle attacks.
Live Threat
Current exploitation, exposure, and threat context
Attackers may find this vulnerability appealing due to its potential for persistent credential theft and lateral movement within a compromised environment, despite requiring authenticated access. The ability to intercept JWTs and manipulate host configurations provides a strong incentive for exploitation, assuming initial access is gained.
- Requires authenticated access first.
- Exploits sandbox misconfiguration.
- Allows credential theft and redirection.
Priority actions
Operational Fix
Recommended remediation, mitigation, and detection steps
Teams should prioritize investigating logs for signs of unauthorized modifications to host configuration files within the Windmill nsjail sandbox environment. Authenticated users with script execution privileges can exploit this to compromise sensitive data or redirect traffic, so focus on detecting any suspicious changes to `/etc/hosts`, `/etc/resolv.conf`, or certificate stores.
- Monitor for unauthorized /etc/hosts modifications.
- Review script execution logs for anomalies.
- Isolate affected worker pods if suspicious activity is confirmed.
