Question

Your team develops prompts for a customer-support chatbot using Azure AI Foundry prompt flow. The lead engineer requires that every prompt change be reviewed before it reaches the production flow, that experimental prompt wording be isolated from the working version, and that the full history of who changed each prompt and why is preserved. The prompt flow files are already stored in a Git repository. Which approach best satisfies these requirements?

Accepted Answer

Develop each experimental prompt on a dedicated feature branch, open a pull request into the main branch, and require peer review approval before merging.

Answer

Keep all prompt edits directly on the main branch and rely on Foundry run history to track which prompt produced each evaluation result.

Answer

Export the prompts as a versioned data asset in the workspace before each change so the asset version numbers form the change history.

Answer

Store each prompt revision as a separate file suffixed with a timestamp in the main branch and delete older files after deployment.