CVE-2026-42076

CVE-2026-42076 is a command injection vulnerability (CWE-78) affecting Evolver, a GEP-powered self-evolving engine for AI agents, in versions prior to 1.69.3. The issue resides in the _extractLLM() function, which constructs a curl command through unsafe string concatenation and executes it via execSync() without proper input sanitization. This allows attackers to inject shell metacharacters into the corpus parameter, leading to arbitrary shell command execution on the server. The vulnerability carries a CVSS v3.1 base score of 9.8 (AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H), indicating critical severity.

The vulnerability is exploitable remotely over the network by unauthenticated attackers with no privileges or user interaction required. An attacker can craft a malicious corpus parameter containing shell metacharacters, such as command separators or backticks, to inject and execute arbitrary commands when the _extractLLM() function processes the input. Successful exploitation grants full remote code execution (RCE) on the affected server, potentially enabling data theft, persistence, or further compromise.

Mitigation is available in Evolver version 1.69.3, which patches the command injection flaw. Security practitioners should upgrade to this version immediately. Official details are provided in the GitHub security advisory (GHSA-j5w5-568x-rq53) and the release notes for v1.69.3.

As a component in AI agent workflows, this vulnerability highlights risks in self-evolving AI engines where unsanitized external inputs can propagate to system-level execution, though no public evidence of real-world exploitation has been reported.