Spec-Driven Development with AI in Eclipse Che

This repository provides a template to demonstrate and experiment with the concept of Specification-Driven Development using AI assistants in a pre-configured Eclipse Che workspace.

The goal is for you to define an application's requirements in the spec.md file and then use AI tools (RA.Aid and Aider) to autonomously generate the code.

How to Use This Template

1. Launch the Workspace

Use the devfile.yaml in this repository to create a new workspace in Eclipse Che or Red Hat OpenShift Dev Spaces. This will automatically pull a pre-configured container image with all the necessary AI tools installed.

2. Configure Your API Key

The AI assistants require an API key from a Large Language Model (LLM) provider to function. RA.Aid and Aider are flexible and support many providers, including Gemini, OpenAI, Anthropic, Groq, and more.

The recommended way to provide your key is by creating a Kubernetes Secret. The workspace is configured to automatically mount this secret as an environment variable.

For example, to use the Gemini API, you would create a secret for the GEMINI_API_KEY environment variable like this:

oc apply -f - <<EOF
kind: Secret
apiVersion: v1
metadata:
  name: gemini-api-key
  labels:
    controller.devfile.io/mount-to-devworkspace: 'true'
    controller.devfile.io/watch-secret: 'true'
  annotations:
    controller.devfile.io/mount-as: env
data:
  GEMINI_API_KEY: [base64 encoded Gemini API key]
type: Opaque
EOF

If you prefer to use a different provider, simply consult the RA.Aid or Aider documentation for the correct environment variable name (e.g., OPENAI_API_KEY) and adapt the secret accordingly.

3. Write Your Specification

Edit the spec.md file to describe the application you want to build. Be as detailed as possible, defining user stories, UI/UX, API endpoints, data models, and technologies.

4. Run the AI Agent

Once you have crafted your specification, run this command in the workspace terminal:

ra-aid --use-aider --cowboy-mode --show-cost --track-cost --msg-file spec.md

The AI will read your specification and begin building your application.

Example Project: URL Shortener

To see a complete, working example, check out the demo-url-shortener branch. It contains the exact spec.md used to generate a full-stack URL shortener and all of the resulting source code.