Sensitive data, such as personally identifiable information and human genetic data, should be protected against unauthorized accesses. When such sensitive data is stored and analyzed in a secure and trusted e-infrastructure, the workflow typically includes uploading raw data to the e-infrastructure, processing and downloading the results. However, accessing sensitive data imposes a significant added complexity due to the security restrictions of the e-infrastructure. This complexity results in additional manual work for researchers to perform data processing especially for use cases where raw data is generated incessantly, and instant result collection is a requirement.Stroll is a universal filesystem-based interface for seamless job submission to high performance computing clusters. Users interact with the cluster through simple read and write operations. Stroll supports simple and composite jobs.This paper describes the client-server architecture of Stroll that enables job submission to isolated sensitive data clusters. Stroll client-server is currently implemented in a pilot project for job submission in the Services for Sensitive Data e-infrastructure at the University of Oslo, and is part of an ongoing effort for building a Nordic federated sensitive data cloud.