ADK SDLC Agent Chain

This project uses the Agent Development Kit (ADK) to create a chain of AI agents that automate the software development lifecycle for any software project.

Architecture

The agent chain consists of the following components:

1. Orchestrator Agent - Coordinates the workflow between all other agents
2. Requirements Agent - Connects to Atlassian/JIRA via MCP to fetch ticket descriptions
3. Implementation Agent - Writes code based on the requirements
4. Testing Agent - Creates unit tests for the implementation
5. PR Agent - Submits code changes to GitHub using GitHub MCP

Getting Started

Prerequisites

• Python 3.9 or higher
• Poetry
• Google Cloud account with Vertex AI enabled
• Atlassian/JIRA account with API access
• GitHub account with access to your target repository

Setup

1. Clone this repository
2. Copy .env.example to .env and fill in your credentials
3. Install dependencies:

poetry install

Running the Agent Chain

Start the agent chain with:

cd adk-sdlc
adk run .

Or use the ADK web interface:

cd adk-sdlc
adk web

Then select the ADK SDLC agent from the dropdown menu in your browser.

Usage

Once the agent is running, you can initiate the workflow by providing a JIRA ticket ID. The agent chain will:

1. Fetch the ticket description from JIRA
2. Generate an implementation based on the requirements
3. Create unit tests for the implementation
4. Submit a pull request to your target repository

Customization

You can customize the behavior of each agent by modifying its prompt in the respective prompt.py file.