Description

OpenChoreo is a complete, open-source developer platform for Kubernetes. Prior to 1.0.4, 1.1.4, and 1.2.0-rc.2, OpenChoreo Workflow Plane templates under samples/getting-started/workflow-templates/ interpolated developer-controlled workflow parameters into shell program text executed through sh -c instead of passing the values through container.env, allowing arbitrary commands to run in workflow pods while affected privileged Podman templates lacked hostUsers: false. This issue is fixed in versions 1.0.4, 1.1.4, and 1.2.0-rc.2.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "8.8",
  "pubDate": "2026-08-13T22:17:28.580Z",
  "pubdate": "2026-08-13T22:17:28.580Z",
  "executiveSummary": "OpenChoreo Workflow Plane suffers from a command injection vulnerability due to insecure interpolation of developer-controlled workflow parameters into shell program text executed via sh -c.\nThis flaw allows malicious actors with the ability to define workflow parameters to execute arbitrary commands within workflow pods.\nAdditionally, affected privileged Podman templates lack the hostUsers: false configuration, increasing the severity of potential container breakout or host-level compromise scenarios.\nThe vulnerability impacts OpenChoreo versions prior to 1.0.4, 1.1.4, and 1.2.0-rc.2.\nSuccessful exploitation grants an attacker arbitrary command execution capabilities within the context of the running workflow container.\nMitigation requires upgrading to the patched versions where input handling and template security are properly enforced.",
  "technicalDetails": "The root cause of the vulnerability resides in the template design within samples/getting-started/workflow-templates/, where developer-controlled workflow parameters are directly interpolated into shell command strings rather than being passed securely through container environment variables (container.env).\nBecause the input is concatenated directly into the shell execution string processed by sh -c, specially crafted input strings can break out of the intended argument context and inject arbitrary shell commands.\nThe attack flow begins when a user supplies malicious workflow parameters containing shell metacharacters or command chaining operators.\nWhen the workflow plane evaluates the template, it substitutes the raw parameter values directly into the shell execution text.\nUpon execution, the shell parses the injected instructions alongside the legitimate commands, leading to arbitrary code execution within the workflow pod.\nFurthermore, the vulnerability is compounded in affected privileged Podman templates that fail to specify hostUsers: false, potentially broadening the attack surface by failing to properly isolate user namespaces.\nThe affected components are the OpenChoreo Workflow Plane templates located under samples/getting-started/workflow-templates/ in versions prior to 1.0.4, 1.1.4, and 1.2.0-rc.2.\nExploitation requires the ability to supply or influence workflow parameters processed by the vulnerable templates.\nPost-exploitation impact includes arbitrary command execution inside the workflow pods, potential lateral movement within the Kubernetes cluster, and system compromise depending on the privileges assigned to the affected pods and missing security contexts such as hostUsers: false."
}