Promptflow is a Python toolset for end-to-end development of LLM-based AI applications, from prototyping to production deployment and monitoring. It includes commands for creating and testing flows, connecting to OpenAI/Azure OpenAI, and evaluating flow quality.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Prompt flow is a suite of development tools designed to streamline the end-to-end development cycle of LLM-based AI applications, from ideation, prototyping, testing, evaluation to production deployment and monitoring. It makes prompt engineering easier and enables you to build LLM apps with production quality.
How it works
- Create executable flows that link LLMs, prompts, Python code and other tools together.
- Debug and iterate flows, including tracing interaction with LLMs.
- Evaluate flow quality and performance with larger datasets.
- Integrate testing and evaluation into CI/CD.
- Deploy flows to a serving platform or into your app's codebase. Optionally use cloud version of Prompt flow in Azure AI.
Getting started
Installation:
pip install promptflow promptflow-tools
Quick Start examples:
- Create a chatbot flow:
pf flow init --flow ./my_chatbot --type chat
- Setup a connection for OpenAI key:
pf connection create --file ./my_chatbot/openai.yaml --set api_key=<your_api_key> --name open_ai_connection
- For Azure OpenAI key:
pf connection create --file ./my_chatbot/azure_openai.yaml --set api_key=<your_api_key> api_base=<your_api_base> --name open_ai_connection
- Test chat flow interactively:
pf flow test --flow ./my_chatbot --interactive
Additional guidance:
- The readme mentions a 15-minute tutorial guiding prompt tuning, batch testing, and evaluation for production readiness.
Recent releases
Latest five releases include versions 1.17.1 (2025-01-09), 1.17.0 (2025-01-06), 1.16.2 (2024-11-25), 1.16.1 (2024-10-08), and 1.16.0 (2024-09-30).
Traction
Stars: 11213 Forks: 1117 Open issues: 70
License
MIT
Data and contributions
The project is in Python, MIT licensed, created 2023-06-30, last push 2026-07-09. Topics include ai, ai-application-development, ai-applications, chatgpt, gpt, llm, prompt, prompt-engineering.