{"id":"CVE-2023-37274","aliases":["EUVD-2023-41185","GHSA-5h38-mgp9-rj5f"],"status":"published","title":"Python code execution sandbox escape in non-docker version in Auto-GPT","description":"Auto-GPT is an experimental open-source application showcasing the capabilities of the GPT-4 language model. When Auto-GPT is executed directly on the host system via the provided run.sh or run.bat files, custom Python code execution is sandboxed using a temporary dedicated docker container which should not have access to any files outside of the Auto-GPT workspace directory.\nBefore v0.4.3, the `execute_python_code` command (introduced in v0.4.1) does not sanitize the `basename` arg before writing LLM-supplied code to a file with an LLM-supplied name. This allows for a path traversal attack that can overwrite any .py file outside the workspace directory by specifying a `basename` such as `../../../main.py`. This can further be abused to achieve arbitrary code execution on the host running Auto-GPT by e.g. overwriting autogpt/main.py which will be executed outside of the docker environment meant to sandbox custom python code execution the next time Auto-GPT is started. The issue has been patched in version 0.4.3. As a workaround, the risk introduced by this vulnerability can be remediated by running Auto-GPT in a virtual machine, or another environment in which damage to files or corruption of the program is not a critical problem.","severity":{"version":"3.1","score":7.6,"vector":"CVSS:3.1/AV:L/AC:H/PR:L/UI:R/S:C/C:H/I:H/A:H","level":"high","assessor":"cna","from":"cve"},"epss":{"day":"2026-11-10","score":0.00352,"percentile":0.29122,"model":"v2026.06.15"},"facts":{"in_kev":false,"known_exploited":false,"public_exploit":false,"fix_available":false},"cwes":["CWE-94"],"published":"2023-07-13T22:33:41Z","modified":"2026-10-28T02:30:53Z","products":[{"vendor":"agpt","product":"autogpt_classic"},{"vendor":"significant-gravitas","product":"auto-gpt"}],"references":[{"url":"https://github.com/Significant-Gravitas/Auto-GPT/security/advisories/GHSA-5h38-mgp9-rj5f","tags":["x_refsource_CONFIRM","x_transferred","Patch","Third Party Advisory","advisory"]},{"url":"https://github.com/Significant-Gravitas/Auto-GPT/pull/4756","tags":["x_refsource_MISC","x_transferred","Patch","Third Party Advisory","fix"]},{"url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2023/37xxx/CVE-2023-37274.json","tags":["advisory"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2023-37274","tags":["advisory"]}],"sources":{"cve":[{"id":"CVE-2023-37274","title":"Python code execution sandbox escape in non-docker version in Auto-GPT","description":"Auto-GPT is an experimental open-source application showcasing the capabilities of the GPT-4 language model. When Auto-GPT is executed directly on the host system via the provided run.sh or run.bat files, custom Python code execution is sandboxed using a temporary dedicated docker container which should not have access to any files outside of the Auto-GPT workspace directory.\nBefore v0.4.3, the `execute_python_code` command (introduced in v0.4.1) does not sanitize the `basename` arg before writing LLM-supplied code to a file with an LLM-supplied name. This allows for a path traversal attack that can overwrite any .py file outside the workspace directory by specifying a `basename` such as `../../../main.py`. This can further be abused to achieve arbitrary code execution on the host running Auto-GPT by e.g. overwriting autogpt/main.py which will be executed outside of the docker environment meant to sandbox custom python code execution the next time Auto-GPT is started. The issue has been patched in version 0.4.3. As a workaround, the risk introduced by this vulnerability can be remediated by running Auto-GPT in a virtual machine, or another environment in which damage to files or corruption of the program is not a critical problem.","status":"published","published":"2023-07-13T22:33:41.29Z","modified":"2024-10-30T15:53:58.317Z","cvss":[{"version":"3.1","score":7.6,"vector":"CVSS:3.1/AV:L/AC:H/PR:L/UI:R/S:C/C:H/I:H/A:H","level":"high","assessor":"cna"}],"cwes":["CWE-94"],"updated_at":"2026-11-02T14:35:16.454633Z"}],"euvd":[{"id":"EUVD-2023-41185","description":"Auto-GPT is an experimental open-source application showcasing the capabilities of the GPT-4 language model. When Auto-GPT is executed directly on the host system via the provided run.sh or run.bat files, custom Python code execution is sandboxed using a temporary dedicated docker container which should not have access to any files outside of the Auto-GPT workspace directory.\nBefore v0.4.3, the `execute_python_code` command (introduced in v0.4.1) does not sanitize the `basename` arg before writing LLM-supplied code to a file with an LLM-supplied name. This allows for a path traversal attack that can overwrite any .py file outside the workspace directory by specifying a `basename` such as `../../../main.py`. This can further be abused to achieve arbitrary code execution on the host running Auto-GPT by e.g. overwriting autogpt/main.py which will be executed outside of the docker environment meant to sandbox custom python code execution the next time Auto-GPT is started. The issue has been patched in version 0.4.3. As a workaround, the risk introduced by this vulnerability can be remediated by running Auto-GPT in a virtual machine, or another environment in which damage to files or corruption of the program is not a critical problem.","status":"published","published":"2023-07-13T22:33:41Z","modified":"2024-10-30T15:53:58Z","cvss":[{"version":"3.1","score":7.6,"vector":"CVSS:3.1/AV:L/AC:H/PR:L/UI:R/S:C/C:H/I:H/A:H","level":"high"}],"updated_at":"2026-11-03T00:05:14.330068Z"}],"nvd":[{"id":"CVE-2023-37274","description":"Auto-GPT is an experimental open-source application showcasing the capabilities of the GPT-4 language model. When Auto-GPT is executed directly on the host system via the provided run.sh or run.bat files, custom Python code execution is sandboxed using a temporary dedicated docker container which should not have access to any files outside of the Auto-GPT workspace directory.\nBefore v0.4.3, the `execute_python_code` command (introduced in v0.4.1) does not sanitize the `basename` arg before writing LLM-supplied code to a file with an LLM-supplied name. This allows for a path traversal attack that can overwrite any .py file outside the workspace directory by specifying a `basename` such as `../../../main.py`. This can further be abused to achieve arbitrary code execution on the host running Auto-GPT by e.g. overwriting autogpt/main.py which will be executed outside of the docker environment meant to sandbox custom python code execution the next time Auto-GPT is started. The issue has been patched in version 0.4.3. As a workaround, the risk introduced by this vulnerability can be remediated by running Auto-GPT in a virtual machine, or another environment in which damage to files or corruption of the program is not a critical problem.","status":"published","published":"2023-07-13T23:15:10.82Z","modified":"2026-06-17T06:07:51.993Z","cvss":[{"version":"3.1","score":7.5,"vector":"CVSS:3.1/AV:L/AC:H/PR:L/UI:R/S:C/C:H/I:H/A:H","level":"high","assessor":"secondary"},{"version":"3.1","score":7.8,"vector":"CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","level":"high","assessor":"primary"}],"cwes":["CWE-94"],"updated_at":"2026-11-02T14:08:15.91037Z"}],"osv":[{"id":"CVE-2023-37274","title":"Python code execution sandbox escape in non-docker version in Auto-GPT","description":"Auto-GPT is an experimental open-source application showcasing the capabilities of the GPT-4 language model. When Auto-GPT is executed directly on the host system via the provided run.sh or run.bat files, custom Python code execution is sandboxed using a temporary dedicated docker container which should not have access to any files outside of the Auto-GPT workspace directory.\nBefore v0.4.3, the `execute_python_code` command (introduced in v0.4.1) does not sanitize the `basename` arg before writing LLM-supplied code to a file with an LLM-supplied name. This allows for a path traversal attack that can overwrite any .py file outside the workspace directory by specifying a `basename` such as `../../../main.py`. This can further be abused to achieve arbitrary code execution on the host running Auto-GPT by e.g. overwriting autogpt/main.py which will be executed outside of the docker environment meant to sandbox custom python code execution the next time Auto-GPT is started. The issue has been patched in version 0.4.3. As a workaround, the risk introduced by this vulnerability can be remediated by running Auto-GPT in a virtual machine, or another environment in which damage to files or corruption of the program is not a critical problem.","status":"published","published":"2023-07-13T22:33:41.29Z","modified":"2026-10-28T02:30:53.85322454Z","cvss":[{"version":"3.1","score":7.5,"vector":"CVSS:3.1/AV:L/AC:H/PR:L/UI:R/S:C/C:H/I:H/A:H","level":"high"}],"cwes":["CWE-94"],"updated_at":"2026-11-02T15:44:32.510144Z"}]}}