Prompt flow
An open-source toolkit for building and evaluating LLM apps locally, with Python workflows, a VS Code designer and optional Azure AI collaboration.
Prompt flow is an MIT-licensed, open-source toolkit for developers who build LLM applications and need to test their behavior before deployment. Its development tools run locally, while an optional cloud version in Azure AI supports team collaboration. Feature development has ended.
A flow connects prompts and LLM calls with Python code and other tools in an executable workflow. That gives developers a way to develop application logic alongside prompts, inspect interactions with the model, and debug problems across the flow. The toolkit includes a Python SDK, a command-line interface and a VS Code extension with a visual flow designer.
Evaluation is a central part of the tool. You can run flows against larger datasets and calculate quality and performance metrics, rather than judge an application only by a few individual responses. Testing and evaluation can also run in CI/CD, so teams can check changes before they reach production.
For deployment, developers can use their chosen serving platform or embed a flow in an existing application's code. The supplied examples include a chatbot that answers questions about PDFs, with evaluation as part of the development process.
Local development doesn't mean all data stays on your machine. The software enables telemetry by default and can send usage information to Microsoft; you can disable it.
