CVE-2026-102734: CWE-693: Protection Mechanism Failure in lmfit asteval

The asteval Python library, used for restricted expression evaluation, had a protection mechanism failure in versions before 1.0.9. Specifically, in asteval/astutils.py, the default Interpreter with numpy enabled exposes a ctypes type that allows memory access via from_address. The safe_getattr function blocks the ctypes module but not reachable ctypes type objects, and attribute assignments use unfiltered setattr rather than safe_setattr. An attacker can craft expressions that derive memory addresses using id, read memory through from_address, and write memory via attribute assignments. This vulnerability enables reading or corrupting arbitrary process memory, escaping the sandbox, executing arbitrary code, disclosing secrets, or crashing the host process. The flaw is addressed in version 1.0.9.

Exploitation of this vulnerability can result in arbitrary memory read and write within the process, leading to sandbox escape, potential arbitrary code execution, disclosure of sensitive information, or denial of service via process crash. The CVSS score is 5.5 (medium severity) with local attack vector, low attack complexity, no privileges required, user interaction required, unchanged scope, no confidentiality or integrity impact, but high availability impact.

Upgrade to asteval version 1.0.9 or later, where this vulnerability is fixed. No other mitigations are indicated or required once the update is applied.