Lesson 10: Git worktree and shipping to production, the last lap

The final lesson of the course. When you finish, you'll know how to let several agents work on one project without stepping on each other using git worktree, and you'll know how to put what you've built on the internet for other people to really use: from a static web page to an automatic link that serves a business process.

The problem: many workers on one site step on each other

In lesson 9 you had several agents running in parallel. But letting them all edit one project together exposes a problem: the files in a project aren't perfectly separate, one file calls another, and when two agents touch the same file they trample each other, which people in the trade call a conflict. It's like two crews renovating the same house: the electricians and the plumbers will sooner or later collide at a shared wall.

Programmers solved this long ago with Git: a version control system that lets you split a project into branches (each branch is a copy for one separate job), and when the job is done you merge it back: the machine compares the two versions and applies the changes to the main one. Git worktree goes one step further, in a way that suits the age of agents: each branch is expanded into its own folder on your machine, sharing history but occupying separate ground. Each agent sits in its own folder, so there's no way left to step on each other.

A real example from the original course: a company website needs 3 more pages (about, contact, services). With one command, the machine creates 3 worktrees, and an agent in each folder builds one page in parallel. When they finish, everything merges back into the main version, with an agent stepping in to settle any conflicts. The secret to it “knowing the way on its own” is exactly what lessons 2 and 4 taught: write the worktree convention into CLAUDE.md (roughly: “this project uses git worktrees for parallel work, one folder per job, never edit another branch's files, merge into the main branch when done”). Say it once, and every later session just follows it.

If you're not a professional software person, you don't need to learn a single Git command. Just remember two ideas: one, there is a mechanism for many agents to work in parallel safely, so just tell the machine you want to use worktrees; two, Git is also a time machine: every version is saved, so if something breaks you can go back to the previous one. That's why you should ask the machine to save a checkpoint (commit) after each stretch that works. It is also your real safety net when you use the broader permission modes from lesson 5.

Going live: from your machine to the internet

All course long, everything you built has sat on your own machine. The last loop is putting it in front of other people. There are two levels:

Level 1: a static web page. An About page or a campaign page that needs no server-side processing can use services like Netlify, Vercel: tell Claude Code “deploy this page to Netlify,” and it handles everything from creating the account and connecting it until you're holding a link to send to customers. It's fast, there's a free plan, and because a static page has no server-side “guts” running, the risk surface is low too.

Level 2: a link that does work (an API). This is the real payoff for automation people: turn a process into a link that runs the process whenever anyone calls it. The service the original course uses is Modal: sign up, get a token, hand it to Claude Code along with the command, and a few minutes later you have a public URL of your own. The demo in the course is a fun link (call it and it tells you “is it someone's birthday today?”), but taken seriously it becomes a door that connects your systems: a new order on the web fires a signal (webhook) into your link, and the link triggers a flow that classifies the order and drafts a reply. Or the lead-filtering skill from lesson 7 gets put on a link, and anyone in the company can open it and fill in a form to make the machine run, with no need to know what Claude Code is. This kind of setup is often surprisingly cheaper than subscriptions to drag-and-drop automation platforms.

Here we repeat the boundary we've stated all course, because it matters most in this very last lesson: a link that runs for fun in-house and a system carrying real customers are two different leagues. Anything that starts to hold customer data, take real money, and have many people depending on it every day, that needs a properly configured server, access control, backups, and someone on call who is accountable. The original course's own instructor, after making good money from these tools, came to the same conclusion: the products he builds with AI are only for internal use and regular clients, not sold at scale on the internet, because he has seen too many security breaches. A smart person uses a sharp knife and knows it is sharp.

Course wrap-up: how far you've come

Looking back at the ten lessons: installing and reading the tool (lesson 1), building a website by talking and CLAUDE.md (lesson 2), the habit of verifying (lesson 3), the .claude folder and rules for writing instructions (lesson 4), permission modes and plan mode with a full app including login and payment (lesson 5), managing context and tokens like managing cost (lesson 6), packaging a workflow into a skill (lesson 7), plugging in outside tools with care (lesson 8), orchestrating sub-agents and agent teams (lesson 9), and running in parallel safely and going live (lesson 10).

You now hold something most people talking about AI out there have never done with their own hands: a complete process from idea to a working product. Our closing advice for the course comes in three sentences. Start with the task you repeat most in your own week and turn it into a skill first. Anything meant for customers and real money should be reviewed by someone who does this for a living. And don't let this knowledge sit idle: automate one more task each week, and three months from now your company will run very differently.

If you want a partner for the “doing it for real for your business” stage, from validating an idea to building a system that holds up under load with someone accountable, visit MONA's in-depth consulting articles on mona.media or call 1900 636 648 directly. As for this free learning series, the next course we are releasing is Advanced Claude Code: hooks, scheduled automation, deeper Chrome DevTools, and architectures for running machines at the scale of real work.

Hands-on to close the course

1. Ask the machine to record the git worktree convention in a project's CLAUDE.md, then give it a job that needs 2-3 parallel branches.
2. Deploy the website from lesson 2 to Netlify and send the link to a friend to open on their phone.
3. Sign up for Modal and build your first link of your own (copying the birthday demo is fine), then call it from a browser.
4. Write down three tasks in your company that deserve an “automatic link,” and score each one's risk against the boundary you've learned.

Going live in Vietnam with MONA Cloud

Type a single command, and fifteen seconds later you get an HTTPS link you can actually touch. It feels very different from watching many people wrestle with server configuration for half a day. With Claude Code working alongside monacloud-mcp, you can easily take a project from your computer to the internet without many intermediate steps. This infrastructure provides a free simulated sandbox environment, and one plus we always stress is that all data stays on servers in Vietnam. Measurements from our internal system on September 5, 2026 show that creating a high-speed Cloud VPS takes only 11 seconds, with latency from the VNPT network in Ho Chi Minh City of about 5 to 15ms, far from the 30 to 50ms you get renting a server in Singapore. Once the system runs steadily, the built-in utility helps you register a custom domain and point traffic to it neatly. Pricing is transparent and billed by the hour, with the smallest plan from 550 VND per hour, paid entirely in Vietnamese dong, plus a 10% VAT invoice each month for your accountant to book. A heads-up: when you stop the system, storage is still charged at 15 VND per GB per hour, so to stop paying completely you need to delete the service yourself. Honest advice: if you are only writing a personal app to try things out, we'd stop you right there. Don't top up for a server, just use up the free sandbox first. Don't rush to burn money.

Integrate login and automatic bank API

Picture this: your website runs smoothly and a customer puts items in the cart, but at checkout they have to message you for your account number, and they get discouraged. Meanwhile you scramble to open your banking app to check whether the money has arrived, manual work that costs effort and is easy to miss. To fix this, we wired the bank API and the automatic payment confirmation service of MONA Pay straight into your workflow. The VietQR code feature, which pre-fills the amount and order code, puts an end to customers mistyping numbers, so money lands directly in your account without waiting for reconciliation. On September 17, 2026, we successfully tested having an AI call our internal top-up tool by itself and show the scan code right in the chat window, so users can top up credits instantly. A few days earlier, on September 15, we had a taste of data mismatch when the security keys between two internal apps didn't match, so money was deducted but the customer's wallet got no credits. Learning hard from that, we now require every command passed back and forth to be signed with HMAC-SHA256, and to be rejected automatically if the delay exceeds 5 minutes, to protect the money flow. This secure payment integration works closely with the MONA Pass system, which automatically issues JWT credentials used across apps from user management to the mailbox, so customers don't have to log in repeatedly. In our view, compared with the user management service from Auth0 that pushes you onto a paid plan once you pass 25,000 monthly active users, our solution is much more economical. The free monthly plan of MONA Pay covers 500 transactions with every feature included, and takes no percentage of the real transaction amount. Our honest advice: once money is involved, don't stitch together software that reads phone notifications to process orders. One wrong bank signal and you lose real money, so hand that job to a dedicated platform. Let the machine handle it.