AMSN revival/ AI Microsoft Network Messenger is a fully AI built MSN messnger collabed by 4 AIs and a code so bad it is so messy to debug, working client version is version 7.5.0324
PS: Dont be like me and create it with your own hands, this is just an experimental test to see whether an LLM can be reliable at patching or just bad at coding altogether
I am very truly sorry for using AI once again, please read the description at the top.
This Project took 2-3 days to make, I originally plan to patch 6.2, but it did not work, so i moved on to 1.0, then 4.5, then 4.7, it all worked, then i took a massive risk to patch 8.1 which i did not accomplish, so next is Windows Live Messenger 14, which i also failed, then i settled to 7.5.0324 which took a few hours of unhealthy AI consumption and finally made it work on 13/09/2026
Again, very sorry for using AI
just tinker with the old Escargot code, it’s right there and it doesn’t make trees melt
I’m sure there’s myriad reasons why using AI to code auth for anything IM related is a bad idea (security reasons off the top of my head, though that’s speaking as someone who flunked one semester of Java and gave up) but most of all it just feels… redundant, when better options already exist. and made by humans too!
Well, i am not going to host it for these reasons:
Hosting is expensive - I cant just host it because hosting is in dollars, and my money is very… small to dollars
Setup - It is relatively easy to set up, but you have to add entries to the hosts file and install a certificate for Trusted Root Certification Authorization
How to host - i have to setup a website for every port, which is already obvious for the debugging of this
Best Options - Escargot/ Human-made & Human Maintained Code leads to better security, better community, and no AI backlash
This will be good for educational, but bad for actual hosting 
hey atleast he’s being honest about it
Fair enough I guess but still 
For 1. I host my servers locally
For 2. I use hosts and a certificate, but even if your proper patch is good, i like the hosts file and certificate better for some reason(Maybe because i want to torture myself from manually doing everything to setup and make the client as original as possible)
For 3. I referenced the “individual website for every port” on the old MSN Messenger
For 4. Real
btw I now know what you are saying about your “still in use Microsoft Domains”. I am fully fully aware due to my server always get a GET/POST request for /RST2.srf, and i know it is live.login.com because they still use it, i cant see it (because of my HOSTS) but uhm yeah
Well yes. RST2 is, to this day, used for Microsoft Account authentication specifically for requesting an access token (hence… Request Security Token :p). Most things built into Windows that uses Microsoft Accounts utilizes it for requesting an access token for auth.
well, either i dont really care about that or I just want MSN Messenger to work
how is it educational? You’re not learning anything, not the protocol, not the web api and not the code itself (since you said it’s unreadable). Building shit for things you’re passionate about is a fundamentally not what you’re doing if you’re chucking 4 AI Agents at it. You’re not getting paid, you’re not benefiting from this so why burn all of these tokens to have a project you can’t even stand on and declare you made it? I’m not saying don’t use AI as a tool to help you understand the protocol or write mundane code like common utility and whatnot (but always triple-check what it’s outputting) but this is just sad, trust me the learning processes and the struggle will feel 100x as rewarding as any vibecoded slop because you can say that you made that shit.
