Generate and Test Code in Isolated Docker Sandboxes

The Code Generation Sandbox is a Docker-integrated example workflow from promptfoo that provides a sandbox environment for evaluating AI-generated code by executing it inside isolated containers. This example demonstrates the patterns documented in promptfoo's sandboxed code evaluation guide.

Use this example when you want to set up a Docker-based sandbox for code evaluation following promptfoo's integration patterns. The Docker containerization provides isolation for code execution. Do NOT use this if you only need static code analysis or linting - the Docker overhead is unnecessary for syntax-only checks. Avoid it when sub-second response times are required, as container spin-up adds latency.

This is an example configuration that you can initialize and customize for your specific code evaluation needs. The exact inputs and outputs depend on how you configure your promptfoo test cases.

This example integrates with Docker for containerized code execution and promptfoo for prompt evaluation workflows. It follows the patterns documented in promptfoo's sandboxed code evaluation guide.

This example is available for anyone using promptfoo who wants to implement Docker-based code evaluation in their workflow.

Initialize the example with:

npx promptfoo@latest init --example integration-docker/code-generation-sandbox
cd integration-docker/code-generation-sandbox

For complete configuration details and usage patterns, consult the promptfoo documentation on sandboxed code evaluations at https://promptfoo.dev/docs/guides/sandboxed-code-evals.