README.md

AI-Assisted Workflow for KCC Resource Development

This guide introduces the Gemini-powered workflow for developing Config Connector resources. This approach accelerates development by automating the implementation details, allowing you to focus on providing the correct inputs and verifying the final results.

Target Audience

This guide is for developers who:

• Have a good understanding of Kubernetes concepts. Here're some basic ideas:
  • Kuberetes Setup
  • CRD and Controller
  • API convention
• Have an in-depth understanding about declarative friendly API. See https://google.aip.dev/news/2020-10.
• Have an in-depth understanding about Kubernetes reconciliation mechanism.
• Have a clear understanding of what Config Connector is.
• Know the differences between Config Connector, Terraform, and DCL.

The Philosophy: You Prompt, Gemini Executes, You Verify

The core of this workflow is a partnership between you and the Gemini CLI.

• Your Role (The Supervisor): Your primary responsibility is to initiate the process with a clear, scenario-based prompt and then to act as a reviewer. You will verify that the code and the behavior (as captured in the golden test logs) are correct and adhere to Config Connector's standards.
• Gemini's Role (The Implementer): Gemini's job is to perform all the hands-on-keyboard work. It will generate APIs, mappers, fuzzers, controllers, mockgcp and test files, and it will run the necessary commands to record test data.

How to Use This Guide

1. Find Your Scenario: Navigate to the scenarios directory and find the document that matches your development task (e.g. adding a new resource, adding a field).
2. Use the Provided Prompt: Each scenario document now begins with a canonical prompt. Copy this prompt and provide it to the Gemini CLI.
3. Understand the Process: The scenario document outlines the steps that Gemini will take. Use this to follow along with its progress.
4. Verify the Results: After Gemini completes the task, use the "How to Verify" section in the scenario document. This section will guide you on which files to review and will link to our deep-dives and api-conventions documents, which serve as the technical reference for what "good" looks like.

The most critical part leading to the success of this Gemini-driven workflow is your ability to triage and discover issues when Gemini gets stuck. This requires some in-depth Kubernetes and Config Connector knowledge.