CVE-2026-123198

Summary

A server-side template injection (SSTI) flaw in LMFDB's render_knowl function allows unauthenticated attackers to inject and execute arbitrary Jinja2 expressions via request parameters, leading to full remote code execution as the web application user.

Exploitability

Exploitation is trivial for unauthenticated attackers due to lack of input sanitization and direct access to Python builtins through crafted query-string or POST data.

Blast radius

Successful exploitation grants full system compromise with the privileges of the web server process, enabling data theft, service disruption, or lateral movement within the infrastructure.

Detection

Look for HTTP requests to /knowledge/render/<ID> containing suspicious Jinja2 syntax (e.g., {{ config.__class__ }}, {{ ''.__class__.__mro__[1].__subclasses__() }}) or attempts to invoke Python builtins like __import__, eval, or os.system in query parameters or POST body; such patterns may appear in access logs or WAF alerts.

Prioritized remediation

Upgrade to LMFDB commit 01829bd or later, which patches the SSTI vulnerability by sanitizing user-supplied input before template rendering.

Analysis generated locally by garnet-coder (no data left the box). AI-assisted — verify against primary sources before acting.