PraisonAI is a multi-agent teams system. Prior to 1.7.2, the codeMode tool in src/praisonai-ts/src/tools/builtins/code-mode.ts executes model-generated JavaScript with new Function() and with(sandbox), while a regular-expression blocklist can be bypassed with Function('return this')() to recover the global object and by constructing the child_process module name dynamically. An attacker who can influence the code argument can access host process capabilities, read or write files, obtain environm

CVE-2026-57141 affects PraisonAI, a multi-agent AI system, prior to version 1.7.2. The vulnerability exists in the codeMode tool which executes model-generated JavaScript using new Function() with a sandbox. The sandbox can be bypassed using Function('return this')() to recover the global object and dynamically constructing the child_process module name to evade regex blocklist protections. An attacker who can influence the code argument passed to this tool can escape the sandbox entirely. Successful exploitation allows reading/writing files, accessing environment credentials, and executing arbitrary OS commands with PraisonAI process privileges. This represents a critical code injection and sandbox escape vulnerability in an AI agent framework. The issue has been patched in version 1.7.2 of PraisonAI.