Overview of prompt flow tools in Azure AI Studio
Important
Items marked (preview) in this article are currently in public preview. This preview is provided without a service-level agreement, and we don't recommend it for production workloads. Certain features might not be supported or might have constrained capabilities. For more information, see Supplemental Terms of Use for Microsoft Azure Previews.
The following table provides an index of tools in prompt flow.
| Tool name | Description | Package name |
|---|---|---|
| LLM | Use large language models (LLM) with the Azure OpenAI Service for tasks such as text completion or chat. | promptflow-tools |
| Prompt | Craft a prompt by using Jinja as the templating language. | promptflow-tools |
| Python | Run Python code. | promptflow-tools |
| Azure OpenAI GPT-4 Turbo with Vision | Use an Azure OpenAI GPT-4 Turbo with Vision model deployment to analyze images and provide textual responses to questions about them. | promptflow-tools |
| Content Safety (Text) | Use Azure AI Content Safety to detect harmful content. | promptflow-tools |
| Embedding | Use Azure OpenAI embedding models to create an embedding vector that represents the input text. | promptflow-tools |
| Serp API | Use Serp API to obtain search results from a specific search engine. | promptflow-tools |
| Index Lookup1 | Search a vector-based query for relevant results using one or more text queries. | promptflow-vectordb |
| Rerank | Rerank documents based on the relevancy to a given query. | promptflow-vectordb |
1 The Index Lookup tool replaces the three deprecated legacy index tools: Vector Index Lookup, Vector DB Lookup, and Faiss Index Lookup. If you have a flow that contains one of those tools, follow the migration steps to upgrade your flow.
Custom tools
To discover more custom tools developed by the open-source community such as Azure AI Language tools, see More custom tools.
If existing tools don't meet your requirements, you can develop your own custom tool and make a tool package.
To install the custom tools, if you're using the automatic compute session, you can readily install the publicly released package by adding the custom tool package name in the
requirements.txtfile in the flow folder. Then select Save and install to start installation. After completion, the custom tools appear in the tool list. If you want to use a local or private feed package, build an image first, and then set up the compute session based on your image. To learn more, see How to create and manage a compute session.
