BuildXL Chat Bot

An AI assistant that can help with BuildXL tasks.

Currently, it is capable of

• answering questions about BuildXL,
• constructing BuildXL command-line arguments (that even BuildXL developers do not know by heart),
• creating a CloudBuild build request, and
• submitting the build request to the CloudBuild service.

This AI assistant is far from being production quality. One can use it as a platform to learn about semantic kernel and prompt design/engineering.

Feel free to contribute to this chat bot by adding more functionalities or features (e.g., adding semantic memory).

Requirements

• .NET >= 8

How to run

• Set/add environment variable OPENAI_SETTINGS, whose value has information regarding the model, end point, and API key. The format of value is UseOpenAI;Model;EndPoint;ApiKey;OrgId, where

  • UseOpenAI: whether to use Open AI or Azure Open AI; true for using Open AI, and false for using Azure Open AI.
  • Model: model to use, e.g., gpt-3.5-turbo.
  • EndPoint: end point; not applicable when using Open AI, so leave it blank.
  • ApiKey: API key.
  • OrgId: organization id; leave it blank if not applicable.

  Examples:

  • true;gpt-3.5-turbo;;sk-12345; - for OpenAI, with gpt-3.5-turbo model and API key sk-12345
  • false;gpt-35-turbo-a;https://1es.openai.azure.com/;5e1ec7ed; - for Azure OpenAI, with gpt-35-turbo-a model, https://1es.openai.azure.com endpoint, and API key 5e1ec7ed

  Note that in both examples the organization id is left blank.

  To set the environment variable:

  • PowerShell:

    $env:OPENAI_SETTINGS='false;gpt-35-turbo-a;https://1es.openai.azure.com/;5e1ec7ed;'

  • CMD:

    set OPENAI_SETTINGS=false;gpt-35-turbo-a;https://1es.openai.azure.com/;5e1ec7ed;

  • Bash:

    export OPENAI_SETTINGS=false;gpt-35-turbo-a;https://1es.openai.azure.com/;5e1ec7ed;

• Run AI assistant:

  dotnet run

Screenshots

References

• BuildXL, Microsoft Build Accelerator
• Learn semantic kernel:
  • Semantic kernel @ Microsoft
  • Semantic kernel @ Microsoft/Github
• Demystifying Retrieval Augmented Generation with .NET