Executive brief
Apache Airflow is an open-source platform used to automate and manage complex data workflows. A security flaw in its management interface allowed authorized users to view sensitive credentials, such as Slack tokens and database passwords, that should have been hidden. This could allow an internal user with limited permissions to gain unauthorized access to external services and databases connected to the Airflow environment.
Technical details
A vulnerability exists in the GET `/api/v2/connections/{connection_id}` REST API endpoint due to an incomplete redaction allowlist (`DEFAULT_SENSITIVE_FIELDS`). Authenticated users with 'Connection-read' permissions can retrieve sensitive JSON blobs in the 'extra' field if the field names (e.g., webhook_url, bearer, dsn) are not explicitly listed for masking. This allows for the exposure of credentials for Slack, HTTP bearer tokens, and database connection strings. The issue is resolved in version 3.2.2 by expanding the redaction list to include common credential field names. As a workaround, operators are encouraged to use external secret backends instead of inlining credentials in Connection objects.
Affected products
- Apache Software Foundation Apache Airflow < 3.2.2
Timeline
- 2026-05-10: other: Remediation pull request submitted
- 2026-05-15: patched: Fix merged into main branch
- 2026-06-01: disclosed: Public advisory and CVE published