Generating code fix suggestion example

This example will walk you through generating code fixes for a Java application that must be migrated to the target technology quarkus. To generate resolutions for issues in the code, we use the Agentic AI mode and the my-model as the large language model (LLM) that you deployed in OpenShift AI.

Procedure

  1. Open the my-Java project in Visual Studio (VS) Code.

  2. Download the Konveyor AI extension from the Konveyor AI release page in GitHub.

  3. Open Command Palette:

    1. Type Ctrl+Shift+P in Windows and Linux systems.

    2. Type Cmd+Shift+P in Mac systems.

  4. Type Preferences: Open Settings (UI) in the Command Palette to open the VS Code settings and select Extensions > Konveyor AI.

  5. Select Gen AI:Agent Mode.

  6. In the Konveyor AI extension, click Open Analysis View.

  7. Type Konveyor: Manage Analysis Profile in the Command Palette to open the analysis profile page.

  8. Configure the following fields:

    1. Profile Name: Type a profile name

    2. Target Technologies: quarkus

    3. Custom Rules: Select custom rules if you want to include them while running the analysis. By default, Konveyor AI enables Use Default Rules for quarkus.

  9. Close the profile manager.

  10. Type Konveyor: Open the Gen AI model provider configuration file in the Command Palette.

  11. Configure the following in the provider-settings file and close it:

        models:
      openshift-example-model: &active
        environment:
          OPENAI_API_KEY: "<Server's OPENAI_API_KEY>"
          CA_BUNDLE: "<Servers CA Bundle path>"
        provider: "ChatOpenAI"
        args:
          model: "my-model"
          configuration:
            baseURL: "https://<serving-name>-<data-science-project-name>.apps.konveyor-ai.example.com/v1"

    You must change the provider-setting configuration if you plan to use a different LLM provider.

  12. Type Konveyor AI: Open Analysis View in the Command Palette.

  13. Click Start to start the Konveyor AI RPC server.

  14. Select the profile you configured.

  15. Click Run Analysis to scan the Java application.

    Konveyor AI identifies the issues in the code.

  16. Click the solutions icon in an issue to request suggestions to resolve the issue.

    Konveyor AI streams the issue description, a preview of the code changes that resolve the issue, and the file(s) in which the changes are to be made.

    You can review the code changes in the editor and accept or reject the changes. If you accept the changes, Konveyor AI creates a new file with the accepted code changes.

  17. Click Continue to allow Konveyor AI to run a follow-up analysis.

    This round of analysis detects lint issues, compilation issues, or diagnostic issues that may have occurred when you accepted the suggested code change.

    Repeat the review and accept or reject the resolutions. Konveyor AI continues to run repeated iterations of scan if you allow until all issues are resolved.