This is a simple project that demonstrates the use of autogen in the context of a Microsoft Teams AI bot. This bot models privacy review helper. The group is built for two purposes currently:
- Threat model validation (validates threat models)
- RAG Search (searches for information using RAG)
- Visualizer Agent (visualizes the threat model)
The basic structure of the autogen group is as follows:
This project uses Teams AI library and autogen.
- Install the Teams Toolkit extension for Visual Studio Code - Link
- Make sure you have conda
- Clone repo
- run
conda env create -f environment.yml
- run
poetry install
- Update .env. Make sure it includes either
OPENAI_KEY
orAZURE_OPENAI_API_KEY
withAZURE_OPENAI_ENDPOINT
. If you look forbuild_llm_config
in the code, you will see how it builds the config for the model. - Open the project in VSCode. Make sure you have the Teams Toolkit (TTK) extension installed.
- Go to the TTK extension, then click "local" debug under environment. This will create a package under appPackage/build folder and start Teams up. It'll also do the installation for your bot on Teams.