Jeremy ADK Software Engineer

Production-grade software engineering for Google Agent Development Kit (ADK) applications. Covers the full lifecycle: architecture design, agent implementation, testing, deployment automation, and multi-agent orchestration.

Overview

This plugin provides an auto-activating skill that acts as a senior ADK software engineer. It helps you build maintainable, testable, and deployable ADK agents following Google Cloud best practices. Whether you are creating a single-agent tool-calling application or a multi-agent orchestration pipeline, this plugin delivers structured code, comprehensive tests, and deployment-ready configurations.

Installation

/plugin install jeremy-adk-software-engineer@claude-code-plugins-plus

Features

• Agent Architecture Design: Single-agent and multi-agent system patterns (Sequential, Parallel, Loop)
• Clean Code Implementation: Modular project structure with proper separation of concerns
• Comprehensive Testing: Unit, integration, and end-to-end test scaffolding with pytest
• Deployment Automation: CI/CD pipelines for Agent Engine, Cloud Run, and GKE
• Tool Engineering: Typed tool interfaces with FunctionTool, input validation, and structured outputs
• Orchestration Patterns: SequentialAgent, ParallelAgent, and custom workflow coordination
• Production Guardrails: Retries with backoff, timeouts, structured logging, and safe error messages
• Multi-Language Support: Python (primary), Java, and Go

Components

| Type | Name | Description | |------|------|-------------| | Skill | adk-engineer (auto-activating) | Engineer production-ready ADK agents and multi-agent systems |

Trigger Phrases

• "Build an ADK agent application"
• "Create production-ready ADK code"
• "Engineer a multi-agent system"
• "Implement ADK agent with tests"
• "Set up ADK development environment"
• "Design ADK agent architecture"

Prerequisites

• Python 3.11+ (or Java/Go for alternative runtimes)
• google-adk package installed (pip install google-adk)
• Google Cloud project with Vertex AI API enabled
• gcloud CLI authenticated with appropriate IAM roles
• pytest for running tests (pip install pytest pytest-cov)

Quick Start

pip install google-adk google-cloud-aiplatform[agent_engines]

Then ask Claude: "Build an ADK agent that [your use case]"

Correct SDK Patterns

# Agent import (correct)
from google.adk.agents import Agent
from google.adk.tools import FunctionTool

# Define a tool as a plain function
def get_weather(city: str) -> dict:
    """Fetch weather for a city."""
    return {"city": city, "temp_c": 22, "condition": "sunny"}

# Create agent with wrapped tools
root_agent = Agent(
    name="weather_agent",
    model="gemini-2.5-flash",
    description="Answers weather questions for any city.",
    instruction="Use the get_weather tool to answer user questions.",
    tools=[FunctionTool(func=get_weather)],
)

adk web  # Opens browser chat at http://localhost:8000

Deployment

# Deployment via SDK
import vertexai
client = vertexai.Client(project="PROJECT_ID", location="us-central1")
remote_agent = client.agent_engines.create(
    agent=app,
    config={
        "requirements": ["google-cloud-aiplatform[agent_engines,adk]"],
        "staging_bucket": "gs://BUCKET",
    },
)

# Deployment via CLI
# adk deploy agent_engine --project=PROJECT --region=REGION agent_module

Note: There is no gcloud CLI for Agent Engine management. Use the Python SDK (vertexai.Client().agent_engines.*) or the adk CLI for deployment.

Typical Project Structure

my-adk-project/
├── src/
│   ├── agents/              # Agent definitions
│   │   ├── __init__.py
│   │   └── main_agent.py
│   ├── tools/               # Custom tool functions
│   │   ├── __init__.py
│   │   └── custom_tools.py
│   ├── orchestrators/       # Multi-agent workflows
│   │   ├── __init__.py
│   │   └── workflows.py
│   └── config/
│       └── settings.py
├── tests/
│   ├── unit/
│   ├── integration/
│   └── e2e/
├── deployment/              # Terraform / K8s configs
├── .github/workflows/       # CI/CD pipelines
├── requirements.txt
├── pyproject.toml
├── Dockerfile
└── README.md

Use Cases

• Greenfield Agent Development: Scaffold a new ADK project from scratch with best practices
• Production Hardening: Add tests, error handling, and monitoring to an existing agent
• Multi-Agent Systems: Design and implement orchestrated agent teams
• CI/CD Pipelines: Generate GitHub Actions workflows for automated testing and deployment
• Code Review Agents: Build agents that lint, review, and analyze code quality

Integration

Works with:

• jeremy-gcp-starter-examples -- ADK sample code from google/adk-samples
• jeremy-vertex-validator -- Validate agents before deployment
• jeremy-vertex-engine -- Inspect deployed agents
• jeremy-adk-orchestrator -- A2A protocol orchestration
• jeremy-genkit-pro -- Combine ADK multi-agent coordination with Genkit flows

Best Practices Enforced

• Security: Never hardcode credentials; use Secret Manager or environment variables
• IAM: Least-privilege service accounts for all deployments
• Testing: Aim for >80% coverage; test happy paths and error cases
• Error Handling: Structured error responses, retry with backoff, graceful degradation
• Code Quality: Type hints, docstrings, PEP 8 compliance
• Observability: Structured logging, Cloud Monitoring, distributed tracing

License

MIT

Support

• ADK Documentation: https://google.github.io/adk-docs/
• Issues: https://github.com/jeremylongshore/claude-code-plugins/issues
• Discussions: https://github.com/jeremylongshore/claude-code-plugins/discussions

Version

2.1.0 (2026) - Full plugin with production agent patterns, testing, and deployment automation