CVE-2026-57141 Details

A remote code execution vulnerability exists in the PraisonAI multi-agent system, specifically in versions prior to 1.7.2. The issue arises within the codeMode tool, located in 'src/praisonai-ts/src/tools/builtins/code-mode.ts'. This tool executes model-generated JavaScript using 'new Function()' and 'with(sandbox)', creating a sandboxed environment. However, the sandbox is not secure, as the 'with' statement does not prevent access to the global object. An attacker can exploit this by bypassing a regular-expression blocklist to access host process capabilities, read or write files, obtain environment credentials, and execute operating-system commands with the same privileges as the PraisonAI process.

Users can update to PraisonAI version 1.7.2 or later, where this vulnerability has been fixed.