MikiMikiMeowMeow; Or, Getting Fancy With AI
MikiMikiMeowMeow; Or, Getting Fancy With AI

Friday • June 6th 2025 • 6:33:50 pm

MikiMikiMeowMeow; Or, Getting Fancy With AI

Friday • June 6th 2025 • 6:33:50 pm

Above all, the how, is you give your AI a script that you review and customize, and it creates the initial code which is similar to a simple wiki that can edit is own code.


Why will anyone will give you money, because you will customize an AI in a way they need.

For example, you will do the research, text programming and maintenance, to create valid pixel art sprite sheets so that game characters can neatly walk.

You create such a service out of curiosity, and you take weeks and months, the person who needs sprites, has no time, they have a game to release.

To them your sprites will be more valuable than you can imagine, and that is why they will give you money.

They give you a little bit of money, so that they can make a bunch of money.


Credit card processing security is a solved problem, basically you pull a transaction id from your payment processor.

If a hacker was to intercept it, which is weird, because it does not need to be saved anywhere.

They can’t do anything with it, it is meant for one purchase, it has an expiration, cancel, and pay.

So the security is done by the the credit card company, your only real worry is refunds.

But we fix that as well, you give people free credits, so that they can test your product.

They only give you money when they love it, it is a low fee, tested, there will be no charge-backs.


They are buying credits on top of the free ones you give them, these extra credits represent a quota limit.

Eventing has a quota limit, you can’t send a gazillion emails, your very ISP won’t let you download a gazillion files.

Ne notion of credits makes the quota limit, transparent, and you have multiple levels so that heavy user can pay more.


Now lets talk about system architecture, there is an API endpoint, but that is boring.

If another user of a pixel world compatible system buys your service, a cute (there is that word again) representative of your system will show up in their lobby.

It will be a system-compatible character that will notify them about quota, answer billing questions, and alert them to system status, very simple.

It is a decoration, on top of a standard way of doing things, that allows their other bots to talk to yours.

So a bot that generates static pixel art characters, can send these images to your bot for converting them to sprite-sheets.

And then your bot will pass them on, to a bot that splits the sprite sheets, into animations and posts them for a human to review.


I do not think that it is bots all the way down, you will need code, and snippets, but AI will take care of it.


So, this is the part, where I briefly mention what the system code should be.

I want this to be a wiki, but you should do your thing as learn more.

I continue building in the same spirit, that had me select pixel art.

I just want to say Thank You, to the code that made me a better programmer.

There will be better ways, and you should discover them.

I would like this to be a MUD, a multi user dungeon, but code comes first.

A self editable wiki, allow me to give it a name.

I call it MikiMikiMeowMeow, the M in the first Miki stands for Map, the second M stands for MUD.

And MeowMeow stands for the fact, that that's what cats do.

The iki, comes from the word Wiki of course, and it means quick, quick to write, iterate, and use.

This should never be exposed to the internet, it must run on your private computer.

The things you sell, need to be hosted on whatever supports node.js, and whatever AI your product uses will be hosted by a company that sells AI.


MikiMikiMeowMeow is a prototype, we shouldn't think about security yet, because the program does not exist yet.

The old way was security first, but in the age of AI, where MikiMikiMeowMeow can be quickly rewritten, lets not worry about hackers.

Beyond stating that this is private software, that should run on your own computer, and not internet.


A self editable wiki, exposes all of its files for editing, including the main file, which if edited wrong, would prevent the wiki from working.

It is just magical, as magical as an infinite loop.

You can always ask AI to modify the code, and prevent the server.mjs from being edited.

Or introduce permissions or user roles, or run every save through a security bot to check for bugs, hacks and bad ideas.

I must remind you and the old programmers reading this, that AI marks a whole new world.

And it going to be a whole new word every few months from now on, the old advice does not work anymore – I think that’s wonderful.


Not only will MikiMikiMeowMeow allow you to edit articles, code, side menu, footer, text description of code, but also transcude them.

Transclusion is the act, sometimes art, of including one page within another.

But most importantly MikiMikiMeowMeow allows you to hold discussions in context of the page you are on.

Especially with multiple artificial intelligence characters, and even trusted humans through their chat avatars on your system.


It is not possible to cover all the features in one text, but they are tagging, routing, history, and bots.

I provide sample implementation of the first three, as it may make for an interesting chat with the AI.

But the bots are too new to fully understand, we can however be sure that server-push is used here.

So you send a command to the MikiMikiMeowMeow, it calls up a remote or local ai and uses server push to deliver the response.

To keep things simple, which bots are on the MikiMikiMeowMeow page you are on, are stored in a related MikiMikiMeowMeow page, which is server push enabled.

If you are on a page named router, router/bots, would contain the AI that is present on the page.

This means that there are sub pages, or special pages, for every page in the MikiMikiMeowMeow.


Finally, I just want to clarify, that deploying a business, or an AI service, does not include MikiMikiMeowMeow code.

But this may not mean that the code relies on huge external modules, we have entered an age, of purpose built apps.

Ones that run without administrative areas and configuration files even, they are programs written by AI for a specific purpose.

And that means, that their source code is a detailed step by step AI prompt, in MikiMikiMeowMeow, so it is just a text document.

I think modern LLM can generate secure and valid code, but it is always good to check.

So another AI, a professional service will need to run a security scan, both externally on your website, and internally on your code.

It is a good idea, whether it is generated or written manually.

Using off the shelf content management systems, and shopping carts, does not always mean that it is safer.

It is usually a massive code-base with external dependencies, so you have to monitor the content management system, the shopping cart, and program libraries.

It won’t be long, before someone tests this, and determines purpose built AI apps are just safer.

The code base is so much smaller that it boggles the mind, supply-chain attack is non-existent as AI writes the modules too.

Asking AI to write secure code, is probably already safer.

You are not asking the AI to vibe code, you are using other AI to write detailed and security minded instructions first.

Artwork Credit