Agentic Development with Claude
💻 code

Agentic Development with Claude

Build with agents without giving up the wheel. From installing Claude Code to writing skills and custom agents, learn a workflow where Claude does the typing and you keep the architecture, the infrastructure, and the standards.

1 chapter8 min readStarted Nov 17, 2026

Welcome to the Agentic Development series.

Week one was about choosing models and writing prompts. This is where Claude starts doing the work, and where your job shifts from typing code to directing and reviewing it.

What you'll learn:

  • How to install and set up Claude Code, in the desktop app and in the terminal
  • How permission modes work, and how much to hand over for each kind of task
  • How to review diffs, steer mid-run, and manage context and usage
  • How skills (SKILL.md files) turn your repeated work into reusable instructions
  • How custom agents work, and where they fit in a real workflow
  • How to keep ownership of architecture, infrastructure, and standards while an agent writes the code

Who this is for:

Developers who want to use agents without losing the thread of their own codebase. People who have tried an agentic tool and want a workflow, not a demo. Engineers who care about review, safety, and being able to explain every change they merge.

The philosophy:

An agent is a co-driver, not the driver. It can read the pace notes, handle the typing, and run the checks. You hold the architecture decisions, the infrastructure decisions, and the coding standards, and you can explain every line you ship. If the tool vanished tomorrow, the project should still make sense to you.

How each chapter works:

Each chapter starts from the official docs, says what is confirmed and what isn't, and shows you how to check it on your own projects. No hype, no promised multipliers.

← All series