Training: Effective development with Claude Code

Claude Code is an advanced agentic development environment in which an agent can read, edit, and run code as a full-fledged member of the team. Unlike ordinary chat assistants, Claude Code lets you work directly with your projects, execute commands in the terminal, edit files, run tests, and make Git commits, all within a single isolated work session. This training will give you a systematic understanding of Claude Code’s capabilities, its architecture, and best practices for using it to speed up and automate development. You’ll learn how to safely and effectively bring Claude Code into your everyday engineering tasks and team workflows.

Goals and objectives

Understand the architecture and capabilities of Claude Code as an agentic development platform;
Learn to use the Claude Code CLI to analyze, edit, and maintain code on real projects;
Master configuring and extending Claude Code with custom slash commands, skills, hooks, and tools;
Learn how to integrate Claude Code safely into team workflows and CI/CD;
Build a practical understanding of the best approaches and patterns for working with context.

Detailed program

Part 1

Generative AI and the Transformer architecture.
How LLMs work and the main stages of their training.
Capabilities and limitations of LLMs.

Part 2

LLMs on the market and how they are classified.
Overview of Anthropic models and the options for accessing them.
Overview of the Claude family of tools.
Comparison of pricing plans for individuals and organizations.

Part 3

The philosophy behind an agentic development platform.
The agentic loop for solving tasks and managing context.
Overview of the Claude Code ecosystem (CLI, desktop, web).
A detailed look at the capabilities of the desktop and web versions of Claude Code.

Part 4

Installing and setting up the Claude Code CLI.
Key capabilities of the CLI version.
Basic commands for working with the CLI.
Tracking usage against limits and the key factors that drive it.

Part 5

Managing sessions.
Remote sessions and moving sessions between interfaces.
Analyzing past sessions to improve how you work with Claude Code.
The permission system and modes for running commands.
Working with checkpoints.

Part 6

Principles of context engineering.
Basic context management techniques.
Configuration scopes.
Managing memory in Claude Code.
Types of memory at the project and user levels.
Using additional memory sources.

Part 7

The MCP protocol and its features.
Configuring additional tools via MCP.
Ways to use MCP and how it compares with CLI tools.
Key MCP servers for development.
Browser automation.

Part 8

The concept of guardrails.
Configuring custom guardrails with hooks.
Types of hooks and how to configure them.
Building various checks into the lifecycle of a task.
Customizing Claude Code with mods.

Part 9

Extracting shared skills.
Using skills flexibly to free up context.
Subagents for delegating tasks.
Built-in subagents and adding custom subagents.
Ways to manage context with subagents.

Part 10

Multi-agent development patterns.
Agent teams and deterministic workflows.
Isolating the agent’s environment with sandboxing.
Working with git worktree for parallel tasks.

Part 11

Configuring Claude Code for a development team.
Developing and distributing plugins.
Practices for optimizing API usage costs.
Implementing code review with Claude Code.

Part 12

The spec-driven development approach.
The main stages and limitations of SDD, and a modern take on it.
Using Superpowers or Mattpocock Skills to extend Claude Code.
Using Spec Kit to extend Claude Code.
Using OpenSpec to extend Claude Code.

Effective development with Claude Code