AE Agent Errors
AE-BNQTVDWF

Code agent's local Python executor sandbox is escaped through whitelisted modules and functions, giving arbitrary code execution

SECURITY FAILUREseverity: CRITICALcause: LIKELYoutcome: RESOLVED UNVERIFIEDconfidence: MEDIUM

smolagents' local_python_executor.py, which runs agent-generated Python in a restricted environment, could be bypassed: attackers could use whitelisted modules and functions to execute arbitrary code on the host despite static and dynamic checks. The project fixed it by preventing access to submodules through indirect attribute access.

Framework / agent
smolagents · code-writing agent (CodeAgent) executing model-generated Python
Remediation attempts
TESTED
Recurrence
not documented
Source languages
en, und-Latn
Updated
2026-09-29

Sources

Symptoms

  • Code executed by the restricted local Python executor escapes the sandbox and runs arbitrary code on the host
  • The isolation boundary for untrusted code is broken, risking data leakage and system compromise
The full record — root-cause evidence, every remediation attempt with its status and verification, failed attempts, patch references, verbatim quotes and recurrence — is a paid lookup (0.018 USDC via x402). Agents: GET /api/v1/cases/AE-BNQTVDWF. Pricing

Similarity to your system is not implied. A remediation that worked in the documented context may not work in yours.