CVE-2026-34538
Unauthorized XCom Data Exposure in Apache Airflow DAGRun Endpoint
Publication date: 2026-04-09
Last updated on: 2026-04-15
Assigner: Apache Software Foundation
Description
Description
CVSS Scores
EPSS Scores
| Probability: | |
| Percentile: |
Meta Information
Affected Vendors & Products
| Vendor | Product | Version / Range |
|---|---|---|
| apache | airflow | From 3.0.0 (inc) to 3.2.0 (exc) |
Helpful Resources
Exploitability
| CWE ID | Description |
|---|---|
| CWE-668 | The product exposes a resource to the wrong control sphere, providing unintended actors with inappropriate access to the resource. |
Attack-Flow Graph
AI Powered Q&A
Can you explain this vulnerability to me?
This vulnerability in Apache Airflow versions 3.0.0 through 3.1.8 involves the DagRun wait REST API endpoint returning XCom result values to users who only have DAG Run read permissions, such as those with the Viewer role.
The issue arises because the endpoint enforces only DagRun read permissions instead of the more restrictive RESOURCE_XCOM permissions required to access XCom data. This means users with limited read-only roles could access sensitive execution results they should not see.
The vulnerability conflicts with the intended FAB RBAC (Role-Based Access Control) model and security documentation, which treat XCom as a separate protected resource and define the Viewer role as read-only without access to sensitive execution results.
The problem was fixed in Apache Airflow 3.2.0 by adding explicit permission checks for RESOURCE_XCOM when the `result` query parameter is present in the DagRun wait endpoint request, ensuring only authorized users can retrieve XCom results.
How can this vulnerability impact me? :
This vulnerability can lead to unauthorized disclosure of sensitive execution data (XCom results) to users who should only have read-only access to DAG Runs.
Users with limited permissions, such as those assigned the Viewer role, could gain access to internal communication data between tasks, potentially exposing confidential information or internal workflow details.
Such unauthorized access could compromise the confidentiality of data processed within Apache Airflow, increasing the risk of data leaks or misuse.
How can this vulnerability be detected on my network or system? Can you suggest some commands?
This vulnerability can be detected by testing the DagRun wait REST API endpoint with the `result` query parameter to see if XCom result values are returned without proper RESOURCE_XCOM permissions.
Specifically, you can attempt to access the endpoint as a user with only DAG Run read permissions (such as the Viewer role) and check if XCom data is returned, which should not happen.
While no explicit commands are provided, a practical approach would be to use curl or similar HTTP clients to send requests to the DagRun wait endpoint with the `result` parameter and observe the response.
- Example curl command to test the endpoint (replace placeholders accordingly):
- curl -H "Authorization: Bearer <token_with_viewer_role>" "https://<airflow-host>/api/v1/dags/<dag_id>/dagRuns/<dag_run_id>/wait?result=true"
If the response includes XCom result data despite the user having only Viewer role permissions, the system is vulnerable.
What immediate steps should I take to mitigate this vulnerability?
The recommended immediate mitigation is to upgrade Apache Airflow to version 3.2.0 or later, where this issue has been fixed.
The fix enforces proper permission checks on the DagRun wait endpoint, requiring RESOURCE_XCOM permissions to access XCom data.
Until the upgrade can be performed, consider restricting access to the DagRun wait endpoint or limiting user roles to prevent unauthorized access to XCom data.
How does this vulnerability affect compliance with common standards and regulations (like GDPR, HIPAA)?:
This vulnerability allows users with only DAG Run read permissions (such as the Viewer role) to access XCom result values, which are considered sensitive execution results. This behavior conflicts with the intended role-based access control (RBAC) model that treats XCom data as a separate protected resource.
Unauthorized access to sensitive execution data could lead to exposure of confidential information, potentially violating data protection requirements in standards and regulations like GDPR and HIPAA that mandate strict access controls and data confidentiality.
By allowing broader access than intended, the vulnerability undermines the principle of least privilege and could impair compliance with these regulations until the issue is resolved by upgrading to Apache Airflow 3.2.0 or later.