A team develops multiple AI applications in Microsoft Foundry that rely on shared prompt templates.

A Git repository provides all three capabilities required: version history for every change with author attribution and timestamp, branching for experimentation, tagging for stable releases, and a single source of truth accessible to all projects via clone or submodule reference. Azure ML datasets (option B) are designed for training data, not text configuration artifacts, and lack per-line diff tracking and review workflows. Embedding prompts in application configuration files (option C) scatters prompt content across multiple applications with no unified view, no shared versioning, and no cross-project reuse. Unstructured Blob Storage with folder organization (option D) provides no version history, no diff tracking, no review workflow, and no native tooling for comparing prompt variants. Git is the definitive answer for centralized tracking, versioning, and reuse of prompt content across Microsoft Foundry projects.

Microsoft Learn Reference Topic: Prompt engineering and version control – Managing prompts in AI applications with Git