{"uuid": "e1c12dd9-c6b8-4272-b887-67745fae7932", "vulnerability_lookup_origin": "1a89b78e-f703-45f3-bb86-59eb712668bd", "author": "9f56dd64-161d-43a6-b9c3-555944290a09", "vulnerability": "CVE-2026-51857", "type": "seen", "source": "https://gist.github.com/Ro1ME/78606e0763520d6519897e90b305d3cd", "content": "# CVE-2026-51857\n\n**Type:** Incorrect Access Control\n**Product:** camel-ai\n**Affected versions:** 0.2.91a1\u20130.2.91a3\n**Affected component:** `camel/interpreters/subprocess_interpreter.py:SubprocessInterpreter.run_file`\n**Reference:** https://github.com/camel-ai/camel/issues/4037\n\n## Description\n\nIn camel-ai camel 0.2.91a1, v0.2.91a2 and v0.2.91a3, CodeExecutionToolkit can run model-produced Python code through SubprocessInterpreter without an approval boundary.\n\n## Discoverer\n\nRo1ME\n", "creation_timestamp": "2026-09-26T04:02:16.000000Z"}