LINE Workflow Automation with Claude Code & MCP

Course overview

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 assigned in LINE and preparing a report for management every morning. Today, AI agents such as Claude Code connected to the LINE Bot MCP Server, LINE's official MCP server, let us give instructions in plain Thai and have the AI carry out the work from start to finish.

The course is built as two connected steps. Day one teaches Claude Code thoroughly for beginners: installing it, giving instructions, controlling permissions with Plan Mode, managing context and CLAUDE.md, connecting MCP to a PostgreSQL database and having AI write automation scripts. Day two puts all of these skills together to build a real LINE automation system: an HR notification bot, a system that captures and summarises group chats through a webhook, and an executive report sent to LINE automatically every morning. Every workshop uses data from a fictional company, and all prompts work with both Claude Code and Codex CLI. (2 days, 6 hours per day, 12 hours in total, Beginner-Intermediate level.)

What you’ll gain

Install, configure and use Claude Code confidently, even with no prior experience of command-line AI agents
Control the AI agent safely with permission modes, Plan Mode, approvals and interrupting mid-task
Manage project context and memory with CLAUDE.md and write prompts that produce accurate results
Connect MCP servers to Claude Code yourself, including PostgreSQL and the LINE Bot MCP Server
Send broadcasts and Flex Messages and create rich menus for internal communication
Design a webhook and database to capture group chats, then have AI summarise meetings and track tasks
Build an executive report from a database, send it to LINE and schedule it to run automatically
Work with due regard for privacy, consent, PDPA and token security

Who this course is for

People new to Claude Code or command-line AI agents who want to start the right way
Developers and IT staff who want to build LINE automation for their organisation
Power users, data analysts, HR or operations staff who use AI chat regularly
Team leads and technology managers planning to adopt AI agents and LINE

Prerequisites

Comfortable using a computer and installing software, and able to type basic terminal commands
Some experience with an AI chat such as Claude or ChatGPT (no Claude Code experience needed)
A paid Claude plan that includes Claude Code, and a LINE account on your phone
Programming and SQL basics help you go further but are not required

Curriculum

Course Details

This course is the 2026 reformatted edition of LINE Workflow Automation with Claude Code & MCP. The core content is still current; it is now organised into 12 topics, with added coverage of Claude Code permission modes and MCP scopes. It runs for 2 days, 6 hours per day (12 hours in total, 09:00-16:00), as workshops built on one fictional company throughout. Beginner-Intermediate level. Learners use their own paid Claude plan and their own LINE Official Account and ngrok accounts, and take home a prompt library, code and a database seed file.

Day 1 Claude Code for Beginners, End to End

Section 1: From AI Chat to AI Agent

How an AI chat differs from an AI agent that actually does the work
Claude Code and Codex CLI as command-line AI agents
An overview of the LINE automation system you will finish on day two
Activity: collect the LINE workflows your organisation uses into a pool of challenges

Section 2: Lab: Getting Started with Claude Code and Permissions

Install Claude Code on Windows, sign in and check the machine is ready
Sessions, project folders and the instruct, review and approve cycle
Create, read and edit files and run commands through the AI
Permission modes: step-by-step approval, Plan Mode, Auto Mode and switching with Shift+Tab
Press Esc to stop mid-task and redirect the work

Section 3: Key Commands, Context and CLAUDE.md

/help, /clear, /compact, /model, /mcp, /init, and picking up work with continue and resume
What context is, why long sessions make the AI lose track, and how to manage it
CLAUDE.md as project memory for your team's rules
Good prompts: goal, scope, requirements and results you can check

Section 4: Workshop: Your First Project with Claude Code

Have AI build a web page summarising the fictional company's data with Chart.js charts
Review the plan in Plan Mode, approve it and iterate until you are satisfied
Create CLAUDE.md and test that its rules apply to the next task
These skills underpin the webhook server and reports on day two

Section 5: Lab: MCP to Connect AI with Company Data

What MCP is, its architecture and ready-to-use MCP servers
Add servers with claude mcp add and choose local, project or user scope
Import 30 days of sample sales data from a seed file
Connect the Postgres MCP server and ask for sales summaries and trends in Thai
Lab: find the unusual sales records hidden in the data

Section 6: Lab: Automated Report Scripts and Codex CLI

Ad hoc questions go live through MCP; routine work uses scheduled scripts
Have AI write a Node.js sales summary script, test it as a dry run and schedule it with Task Scheduler
Configure MCP in Codex CLI and run the same prompts to compare results
Prepare for day two: a LINE Developers account, a test group chat and ngrok

Day 2 Building a Complete LINE Automation System

Section 7: The LINE Platform for Organisations

What LINE Official Accounts and the Messaging API can do, and the message quota of each plan
A key limitation: the Messaging API cannot read past chat history
The overall architecture: inbound webhook, PostgreSQL, Postgres MCP for reading and LINE Bot MCP for sending
Issuing a channel access token and keeping it secure

Section 8: Lab: LINE OA and an HR Notification Bot

Create a provider and Messaging API channel, then configure the webhook and allow the bot to join groups
Connect the LINE Bot MCP Server to Claude Code and send a first message with a Thai prompt
Play the HR role: individual meeting notices, broadcast announcements and checking the message quota
Design a Flex Message for an internal job posting with buttons
Create a rich menu for employee services

Section 9: Lab: A Group Chat Recorder

Have Claude Code build a webhook server with Node.js, Express and @line/bot-sdk
Verify signatures, deduplicate LINE redeliveries and always reply with 200
Store group messages in PostgreSQL
Expose the server over HTTPS with ngrok and set the webhook URL
Lab: team members chat in a real group and check the data in the database

Section 10: Lab: AI Meeting Summaries and Task Tracking

Have AI read chats from the database through MCP and summarise what was discussed
Find appointments and tasks people agreed to, and build a task tracker
Post the meeting summary and open tasks back to the group as a Flex Message
Check the summary is accurate before every send

Section 11: Workshop: Executive Daily Report

Extend the day-one report script to pull yesterday's sales from the database
Build an executive report Flex Message with alerts when unusual records are found
Test as a dry run, check the JSON, then send it to the group for real
Schedule it to run automatically every morning at 08:00 with Task Scheduler

Section 12: Group Challenge, Governance and Privacy

Each team picks a real challenge from the pool, has the AI agent build a prototype and presents it
Get consent before storing group chats, use read-only database users and follow PDPA
Sensitive data that must not leave the organisation, and looking after tokens
Next steps: alerts for unusual events only, your own MCP server and moving to the cloud

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

Join the waiting list Ask about rounds

For organisations — in-house / private

Tailor the content to your team’s tools and projects
Your dates, at your office or live online
Quotation with tax ID for procurement

Corporate training quote

Instructors

Samit KoyomFounder & CEO, IT Genius Engineering

Frequently asked questions

What is Claude Code and how does it differ from an ordinary AI chat?

Claude Code is Anthropic's AI agent that works inside a project folder on your machine, reading and editing files, running commands and connecting to external systems through MCP, whereas an AI chat only replies with text for you to act on. Claude Code offers permission modes, such as asking for approval before editing files or running commands, and Plan Mode so you can review its plan first, and it can be stopped mid-task, so your role becomes directing and checking the work.

How do you add an MCP server to Claude Code?

Run claude mcp add followed by a name of your choice and the command or URL for the server, passing required values, such as a database connection string or a LINE access token, as environment variables. Choose a scope: local for you in this project, project to share it with the team through a .mcp.json file, or user for every project on your machine. Then type /mcp in Claude Code to check the connection and see the available tools.

Why can't a LINE bot read past group messages, and how should messages be stored?

The LINE Messaging API offers no way to fetch chat history. A bot only receives messages as events through its webhook after it has been added to a group. The approach is therefore to run a webhook server that verifies LINE's signature and saves each message to a database such as PostgreSQL straight away, then have AI read from the database to produce summaries. Group members must be informed and give consent before messages are stored.

Who is the LINE Workflow Automation with Claude Code & MCP course for, and what should I know beforehand?

It suits people new to Claude Code or command-line AI agents, developers and IT staff who want to automate LINE for their organisation, HR or operations staff who use AI chat regularly, and team leaders who want to see what is possible. You should be able to install software and type basic terminal commands, have used an AI chat, and use LINE on your phone. Programming or SQL knowledge helps but is not required.

What will I be able to do after the LINE Workflow Automation with Claude Code & MCP course?

You will be able to use Claude Code confidently, including permission modes, Plan Mode, CLAUDE.md and context management, connect the Postgres MCP server to analyse data in plain language, use the LINE Bot MCP Server to send messages, Flex Messages and rich menus, build a group-chat summary system through a webhook, and send executive reports to LINE automatically with Task Scheduler. 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 LINE Workflow Automation with Claude Code course?

Bring a Windows laptop with permission to install software, a paid Claude plan that includes Claude Code, and a personal LINE account on your phone, since you will receive messages and test the bot for real. In class you install Claude Code and set up a LINE Official Account, the Messaging API and ngrok with your own accounts. Practice data comes from a fictional company, with a prepared database seed file provided.