Claude Code Roadmap 12 Units is a hands-on workshop in using Claude Code as an AI coding agent, from the CLI, CLAUDE.md, Plan Mode and Git workflow to MCP, Claude Skills, automation, CI/CD, parallel subagents and plugins. It suits developers, DevOps engineers and tech leads who already code, and you leave with a multi-agent AI coding workflow built in the capstone.
Claude Code is not merely a tool for asking an AI to write code. It acts as an AI coding agent that understands project structure, reads files, edits source code, runs commands, tests the system and works alongside a real development workflow. Anyone still using it one question at a time, copying snippets back into the editor, is capturing only a fraction of what it can do. This course walks learners through 12 units in order: the first Claude Code CLI session, building project context with a CLAUDE.md file, using Plan Mode to separate planning from editing, running the
full agentic loop, integrating with Git from commit through pull request, connecting external tools via the Model Context Protocol, building Claude Skills for repeated work, running headless in automation and CI/CD pipelines, distributing work across parallel subagents, and customizing the environment with plugins for a whole team. It closes with a capstone project that assembles everything into a single multi-agent AI coding workflow. Three major project workshops anchor the course at units 3, 6 and 12, with a 30% lecture and 70% workshop split. (5 days, 6 hours per day, 30 hours in total, Beginner to Intermediate level.)
What you’ll gain
Install, configure and operate the Claude Code CLI from the terminal correctly and safely
Build project context with a CLAUDE.md file so the agent follows project and team standards
Understand Plan Mode and the agentic loop, and know when to take back control
Integrate Claude Code into the version control workflow from code change to pull request
Extend capability through the Model Context Protocol to reach databases, APIs and external systems
Build Claude Skills that turn repeatedly explained knowledge into a permanent capability
Run Claude Code headless and connect it to automation workflows and CI/CD pipelines
Design multi-agent workflows with parallel subagents and synthesize results into one solution
Customize the team environment with plugins, custom commands and shared configuration
Who this course is for
Developers and programmers who want to adopt Claude Code systematically
Full stack, backend and frontend developers interested in AI-assisted development
DevOps engineers who want to wire AI into their automation workflow
Tech leads bringing an AI coding workflow to their team
Anyone using AI for code who wants to move from ad-hoc prompting to an agentic workflow
Software teams adapting Claude Code to their organization workflow
Prerequisites
Programming experience in at least one language and the ability to read a project structure
Basic command line and terminal skills
Basic Git knowledge such as commit, branch and pull request
A codebase or sample project of your own to work with during the course
No prior Claude Code experience needed; unit 1 starts from installation
Curriculum
Course Details
A 5-day course, 6 hours per day (30 hours in total), taught hands-on and project-based with a 30% lecture and 70% workshop split. Beginner to Intermediate level. Learners bring a laptop with Node.js and Git installed, an account for Claude Code, and their own codebase to work with across all five days. The course is divided into 12 units that progress from fundamentals to real application, with three major project workshops at units 3, 6 and 12. It focuses on using Claude Code as an AI coding agent and on designing AI coding workflows; it is not a course on any particular programming language or framework.
Day 1Getting Started and Project Context (Units 1-3)
Unit 1: Getting Started with the Claude Code CLI
Installing and configuring the Claude Code CLI and working through the command line
Directing the agent in natural language and delegating coding tasks
Understanding how Claude Code reads a codebase, edits files and runs commands
Lab: start using Claude Code on a real project and delegate a basic programming task
Unit 2: Building Project Context with CLAUDE.md
What a CLAUDE.md file is and how to create one for your own project
Defining coding conventions, project architecture and codebase standards
Recording the terminal commands you use regularly and your development approach
Lab: write a CLAUDE.md that removes repeated prompting and keeps context consistent
Unit 3: Project 1, Setting Up Claude Code on a Real Project
Connecting Claude Code to your codebase and having it explore the project structure
Creating CLAUDE.md with coding conventions, architecture and project context
Delegating a real development task to Claude Code
Workshop: deliver a project fully configured for Claude Code, ready for day two
Day 2Agentic Workflow and Git Integration (Units 4-6)
Unit 4: Plan Mode and the Agentic Loop
Plan Mode: analyzing requirements, analyzing the codebase and locating relevant files
Proposing architecture and an implementation plan, and assessing impact before coding
The full agentic loop: analyze, plan, implement, test, fix and verify
Lab: run the agentic loop on a real task and decide when to take back control
Unit 5: Using Claude Code with Git
Checking git status, analyzing code changes and staging files with Claude Code
Creating commits and having the agent write meaningful commit messages
Managing branches for features or bug fixes and reviewing code before committing
Lab: prepare and open a pull request through a Claude Code assisted workflow
Unit 6: Project 2, Delivering a Feature End to End
The full workflow: requirement, plan with Claude, implement, test, fix, commit, pull request
Analyzing a feature or bug and planning the implementation with Claude
Editing source code, adding or adjusting tests and verifying the result
Workshop: submit a pull request that is genuinely ready for team review
Day 3External Tools and Reusable Skills (Units 7-8)
Unit 7: Connecting Claude Code to External Tools with MCP
Understanding the Model Context Protocol and the structure of an MCP server
Configuring an MCP server and connecting Claude Code to external tools
Connecting databases, APIs and external data sources, and calling tools from Claude Code
The architecture from Claude Code through MCP to external tools, APIs, databases and services
Permission and security considerations when giving an AI agent access to external systems
Lab: configure an MCP server and connect Claude Code to a real database or API
Unit 8: Building Claude Skills for Repeated Work
What Claude Skills are and how a skill folder is structured
Writing instructions, defining constraints and stating best practices
Packaging workflow and domain knowledge into something reusable
Applied examples: code review, API development, database migration and testing
Further examples: documentation, security checks, deployment and organizational coding standards
Lab: build at least one custom skill for your own recurring work
Day 4Automation, Multi-Agent and Customization (Units 9-11)
Unit 9: Automating Claude Code
Headless and non-interactive operation, and invoking Claude Code from a script
Triggering work automatically and integrating with git hooks
Wiring Claude Code into a CI/CD pipeline: push, pipeline, analyze, test, review, result
Lab: automate code review, test generation or documentation
Unit 10: Parallel Subagents and Multi-Agent Workflow
The parallel subagent model and decomposing a problem into subtasks
Assigning tasks to several agents, for example frontend, backend, test and security
Running work in parallel and gathering the results back into one solution
Lab: split a large task across subagents running concurrently and synthesize the output
Unit 11: Plugins and Customizing the Claude Code Workflow
What Claude Code plugins are, installing them and exploring available extensions
Building custom commands, custom tools and workflow extensions
Sharing configuration across the team and packaging a workflow for distribution
Lab: set up a team environment with the same coding standards, commands, skills and tools
Day 5Capstone, Building a Multi-Agent AI Coding Workflow (Unit 12)
Unit 12: Project 3, Building a Multi-Agent AI Coding Workflow
Assembling every unit: CLI, CLAUDE.md, Plan Mode, the agentic loop and Git
Extending it with MCP, skills, automation, parallel subagents and plugins in one project
Designing a main agent that analyzes and plans, then distributes work to subagents
Assigning roles across coding, testing, review and documentation agents
Connecting external tools through MCP and consolidating the results for verification
Carrying through to test, verify, git commit and pull request
Automating parts of the flow to remove manual developer work
Connecting this workflow to the software development process your team already runs
How to grow it into a workflow used for real work in the organization
Workshop: present the multi-agent AI coding workflow you built and review the design together
Schedule & training options
For individuals — public rounds
No public rounds are open right now. Join the waiting list and we will contact you first when the next round opens, or ask us on LINE. Or call 02-570-8449 or 088-807-9770
CLAUDE.md is a file Claude Code reads automatically when it starts work in a project, holding context you do not want to repeat every time, such as the project structure and architecture, coding conventions, the commands used to build, test and lint, and team agreements. Keep it concise and update it as the project changes. The /init command has Claude Code draft a first version from the codebase for you to refine.
When should you use Plan Mode in Claude Code?
Plan Mode has Claude Code analyse the requirement, explore the codebase and propose a plan without editing any files. It suits changes that touch many files, tasks where the right approach is unclear, and unfamiliar code. Once you have reviewed and adjusted the plan, let it implement. This reduces work going in the wrong direction and keeps architectural decisions with you. Switch into Plan Mode with Shift+Tab.
How do you connect Claude Code to an external database or API through MCP?
Start by choosing or building an MCP server for the system you need, such as a database or an API. Add it to Claude Code with the claude mcp add command or define it in the project's .mcp.json file, then check the connection with the /mcp command. Once connected, Claude Code can call that server's tools while it works. Grant only the access needed, for example a read-only account for a production database.
Who is the Claude Code Roadmap 12 Units course for, and what should I know beforehand?
It suits full stack, backend and frontend developers who want to use Claude Code systematically, DevOps engineers who want to bring AI into automation workflows, and tech leads introducing AI coding workflows to their teams. You should have programming experience in at least one language, be comfortable in the terminal and understand Git basics such as commits, branches and pull requests. No prior Claude Code experience is needed.
What will I be able to do after the Claude Code Roadmap 12 Units course?
You will be able to install and use the Claude Code CLI safely, set project context with CLAUDE.md, use Plan Mode and the agentic loop to take a feature through to a pull request, connect databases and APIs through MCP, build Claude Skills for repeated work, run Claude Code headless in CI/CD, design multi-agent workflows with parallel subagents, and set up plugins and custom commands for your team. Participants who complete the training receive a Certificate of Completion with its own number, which can be verified online.
What do I need to prepare for the Claude Code Roadmap 12 Units course?
Bring a codebase or sample project you want to try with Claude Code, as the workshops have you set up CLAUDE.md, deliver a feature and open a pull request on real code. You should also be comfortable with the terminal and basic Git commands. Installing Claude Code is covered in the first unit, so you do not need to set it up beforehand.
Almost every Thai organisation uses LINE as its main communication channel, from team group chats to LINE Official Accounts, yet that communication is still almost entirely manual: posting announcements, summarising meetings from group chats, chasing tasks…
Today many office workers do repetitive work across dozens of Excel files, wait months for the IT team to build small tools, or have good ideas they want to try but hit the wall of not knowing how to code. The arrival of AI coding agents like Claude Code…
