The LLM Provider panel allows you to configure the application to use your LLM account to generate recommended fixes for static analysis violations. Functionality has been tested on GPT-4o and GPT-4o-mini from OpenAI.


Note:

  • LLM integration in Visual Studio is supported in VS 2019 and later.
  • The functionality does not support VB.NET.
Prerequisites:

  • The "LLM Integration" feature must be enabled in your license.
  • Access to your OpenAI, Azure OpenAI or another LLM provider account is required.
    Note: Only LLM providers with a chat completions endpoint compatible with the OpenAI REST API are supported.

Follow these steps to enable the LLM functionality:

  1. In your IDE, click Parasoft in the menu bar and choose Options.

  2. Select LLM Provider. 
  3. Set the following preferences:

  • Enable: Enables generating AI-recommended fixes for static analysis violations.

  • Provider: Choose between OpenAI, Azure OpenAI or Other.
  • Configuration (OpenAI)
    • API key: Your OpenAI token.
    • Organization ID: The Organization ID you want to use when accessing the OpenAI client. Optional. If no Organization ID is supplied, the default organization ID for your account will be used.
    • Test Connection: Click to test the connection to OpenAI.
    • Model > Chat: The name of your OpenAI model, for example gpt-4o.
    • Model > Embedding: (Optional) The name of your OpenAI embedding model.
  • Configuration (Azure OpenAI)
    • Resource Name: The Azure resource that contains your deployed Azure OpenAI model.
    • API key: Your Azure OpenAI token.
    • Deployment ID > Chat: The deployment name of your Azure OpenAI model.
    • Deployment ID > Embedding: (Optional) The deployment name of your Azure OpenAI embedding model.
    • Test Connection: Click to test the connection to Azure OpenAI.
  • Configuration (Other)
    • Base URL: The URL of the location where your LLM model is deployed.
    • API key / Access token: Your LLM provider token.
    • Model > Chat: The name of your LLM model.
    • Model > Embedding: (Optional) The name of your LLM embedding model.
    • Test Connection: Click to test the connection to the LLM provider.
LLMs may produce inaccurate information, which may affect product features dependent on LLMs.

  • No labels

Need assistance? Visit our support page