Judge0

Judge0 is an open-source, self-hosted online code execution system designed to run untrusted code in a sandboxed environment. It provides a backend infrastructure that allows other applications to compile and execute source code across a wide variety of programming languages without risking the security of the host machine.

The software is deployed as a server and is primarily accessed via a JSON HTTP API or an official Python SDK. Common use cases include building AI agents that generate and test code, competitive programming platforms, e-learning tools, candidate assessment systems for recruitment, and online integrated development environments. By providing a standardized way to handle code execution, it removes the need for developers to build their own secure execution environments from scratch.

Key features

• Sandboxed compilation and execution of untrusted code
• Support for over 90 programming languages
• Simple HTTP JSON API for application integration
• Official Python SDK
• Execution of multi-file programs and projects
• Custom user-defined compiler options and command-line arguments
• Configurable time and memory limits
• Webhooks for HTTP callbacks
• Detailed execution results

The system uses a modular architecture designed for scalability to handle high loads. It is built using Ruby on Rails and leverages Docker and Isolate for its sandboxing capabilities. This architecture allows the system to be deployed on various infrastructures while maintaining a consistent API for the client. Users can choose between two versions, Judge0 CE and Judge0 Extra CE, which differ primarily in the specific set of supported languages and frameworks available for execution.

Judge0 serves as a specialized infrastructure component for developers and organizations requiring a secure, scalable way to execute arbitrary code for humans or AI agents.