How I built an AI agent with Claude Code in 20 minutes
A complete walkthrough: from zero to a working agent using only the Claude Code CLI. No frameworks, no magic.
In-depth guides on Claude API, Claude Code, and AI-powered workflows. No fluff. Only what works.
import anthropic
client = anthropic.Anthropic()
response = client.messages.create(
model="claude-sonnet-4-6",
max_tokens=1024,
tools=[get_weather, search_web],
messages=[{
"role": "user",
"content": user_input
}]
)
# Claude decides which tool to call
if response.stop_reason == "tool_use":
result = handle_tool(response)
send_result_back(result)▋A complete walkthrough: from zero to a working agent using only the Claude Code CLI. No frameworks, no magic.
How to write an effective CLAUDE.md that makes Claude Code more useful and predictable in your project. Structure, examples, and common mistakes.
The most underused feature in Anthropic's API. How it works, when to use it, and how much I saved.
Model Context Protocol explained with real code. From zero to a working TypeScript server with the tools you need.
Extended thinking lets Claude show its reasoning process before responding. Practical guide to enabling it and getting the most from it on complex tasks.
How to design multi-agent systems with Claude that don't spiral out of control. Real coordination and handoff patterns.
No empty theory. These are the techniques that changed results in real projects with Claude.
What I changed, what I kept, and what I learned using Claude Code as my primary development tool.
How to use hooks in Claude Code to run commands before or after each agent action. Full control over what Claude does in your project.
How to implement function calling in Claude API. From tool definition to handling responses in production.
"Roi helped us integrate Claude API into our product in record time. The cost savings with prompt caching were immediate."
"The Claude Code workshop transformed how our dev team works. 3x productivity in the first weeks."
"The multi-agent system Roi designed reduced our content pipeline processing time from 4 hours to 20 minutes."