IBM DataStage on Cloud Pak for Data Path Traversal Arbitrary File Creation

Technical Analysis

The vulnerability stems from insufficient validation of file paths contained within archive files (e.g., ZIP/TAR) processed by IBM DataStage during extraction routines, a classic zip-slip pattern where crafted entry names such as '../../etc/cron.d/malicious' traverse outside the intended extraction directory. An authenticated remote attacker with access to the DataStage job import/upload functionality could exploit this to create or overwrite arbitrary files on the host, potentially leading to remote code execution, configuration tampering, or persistence via scheduled tasks. The CVSS score of 9.1 reflects high impact on integrity and availability with relatively low attack complexity, though authentication is a prerequisite. Organizations running Cloud Pak for Data often use DataStage pipelines to prepare and transform data feeding into RAG pipelines and AI agent data stores; a compromised host in this data pipeline could allow attackers to poison training/inference data, inject malicious files into agent tool directories, or pivot to credential stores used by connected AI agent frameworks.

Affected Systems

IBM DataStage running on IBM Cloud Pak for Data version 5.4.0.0; potentially other Cloud Pak for Data-integrated DataStage deployments using the same archive extraction component.

Indicators of Compromise

• No specific IOCs published; monitor for anomalous archive uploads containing path traversal sequences (e.g., '../', '..\\') in DataStage job/asset import logs.

Remediation Steps

1. 1

   Apply Vendor Patch

   Upgrade IBM DataStage on Cloud Pak for Data to the fixed version specified in IBM's official security bulletin once released.

2. 2

   Restrict Authenticated Access

   Limit which authenticated users/roles can import or upload archive files into DataStage to reduce the attack surface.

3. 3

   Input Validation Monitoring

   Deploy monitoring/WAF rules to detect and block archive uploads containing directory traversal sequences.

4. 4

   File Integrity Monitoring

   Enable file integrity monitoring on DataStage host directories to detect unauthorized file creation or modification.

5. 5

   Isolate Data Pipeline Hosts

   Segment DataStage and Cloud Pak for Data hosts from systems hosting AI agent credentials, RAG vector stores, or model-serving infrastructure to limit lateral movement impact.