When the service is configured to listen on non-loopback interfaces, an attacker can gain unauthorized control over agent instances. Successful exploitation enables an attacker to rewrite configuration files, extract sensitive provider credentials, manage cron jobs, and manipulate MCP (Model Context Protocol) servers. Given the critical nature of these functions, this vulnerability allows for complete compromise of the affected NullHub instance and potential lateral movement into integrated systems. Organizations should restrict management API access to loopback interfaces as an immediate mitigation until patching is performed.

Impact

Successful exploitation results in full unauthorized control over NullHub agent instances. Attackers can exfiltrate provider credentials, establish persistence via modified cron jobs, and execute arbitrary commands through malicious MCP server configurations. This impacts any organization using NullHub for automation or agent management, particularly those exposing the management API to wider network segments.

Recommendation

• Patch NullHub to a version beyond 2026.5.29 immediately.
• Restrict access to the NullHub management API (/api/*) to local loopback interfaces only.
• Monitor webserver logs for HTTP requests to /api/* paths originating from external IP addresses that lack 'Authorization' headers.