Executive brief
Kedro is an open-source Python framework used by data scientists to create reproducible and maintainable data pipelines. A security flaw in version 1.2.0 allows an attacker to bypass directory restrictions by providing a specially crafted version name for a dataset. This could lead to unauthorized access to sensitive files on the system, data corruption, or the exposure of data belonging to other projects or users.
Technical details
A path traversal vulnerability exists in Kedro 1.2.0 due to insufficient sanitization of user-supplied version strings. The `_get_versioned_path()` method in `kedro/io/core.py` directly interpolates these strings into filesystem paths. An attacker with local access can exploit this via the API or the CLI's `--load-versions` parameter (handled by `_split_load_versions()` in `kedro/framework/cli/utils.py`). By using traversal sequences like '../', an attacker can escape the versioned dataset directory to read or write files elsewhere on the filesystem, potentially leading to data poisoning or cross-tenant data access in automated environments.
Affected products
- Kedro Kedro 1.2.0
Timeline
- 2026-06-12: disclosed
- 2026-06-12: advisory