Ninth devlog

I’m finally done with polishing the project up,making a website for it, building it for different platforms,writing the readme and alot of other stuff

and after so much work the web port is finally live too

https://melee.hexdump0.pw/

Eighth devlog

And yep I’ve just been fixing up more crashes, making the website, creating a custom menu item and a custom intro cutscene

Seventh devlog

And I’m finally happy now, I manged to fix most if all of the crashes (thank you claude ) and I optimized the hell out of the render to make it even more playable

And big news! The web port is working too , this took so much time

and yep, the port is 90% of the way done just need to add some qol mods, design some graphics and we are done

Sixth devlog

This is literal hell, I chose to read the disc data as little endian and write converters and walkers for the game

Turns out doing that is not so easy, there is about ~100 different unique crashes that happen (100 is the number of ones I could actually find ) and each one requires a whole lot of time and effort to fix manually

So yea the 10 hours were spent grinding away using my LLMs to get it down to ~20 crashes, there is still alot of work that needs to be done for the 20 crashes and even more for the ones I have found yet

Hopefully by the next devlog I would actually be happy

Fifth devlog

So more bug fixes! There was (and still is) alot of bugs and crashs coz of stupid endianess and layout, But I think I managed to fix a good chunk of them and the game is actually quite playable now excluding it panicing every now and them

Fourth devlog

And we have the full game (almost!), It took alot of effort and time to get here, but now the whole game runs natively

There is still alot of bugs, crashes and other stuff that has to be fixed before it actually playable tho

Hopefully I would have a functional port by this week

Third devlog

Crazy! Crazy! progress so far

I have gotten to actual decompiled game logic to run
It was not easy I spent a whole day fighting with endianness, finding little gamecube quirks and whole alot of other stuff

But alas, we have a match running
All thats left is the game ui and polish

Second devlog

So I did a whole bunch of work, this included kinda throwing out my current progress as I realized hand porting 100k lines of code isn’t really viable

The new architecture is to compile the actual decomped codebase and write shims/wrapper around it for the non portable stuff (dics,audio,render engine,etc)

I rewrote the entire renderer for this new architecture and manged to get stages rendering

First devlog

(p.s I am really bad at devlogging consistently )

Alright so Super Smash Bros Melee for the game cube was fully decompiled last week
and its the perfect time to port it to modern computers and to also benchmark LLMs on what they can achieve

So far I was able to get a model/animation viewer working!

also nitendo please dont vaporizer me thank you

Second devlog (and kinda the final one )

I kinda forgot that I had made a Stardance project for this

But I did do alot of stuff like

• Planed out the economy and the general abstract guidelines
• Finished making the website
• Made a launch video (ik its pretty bad )

But its pretty much ready for RSVP (and for a sponser too )

First devlog

I did quite a bit vibing work , decided to go with a ascii kinda themed website, and this is what I managed to whip up

Ik this devlog is quite a bit late but I worked on ALOT of stuff

Here is a quick list of everything I did

• Worked on choosing the color pallet,fonts and all the other fun stuff
• Made a logo for Pandora
• Made a design system for Pandora
• Made a cool looking ascii background thingy
• Made the hero section of the website part
• Worked on some dashboard mockups

So if yall havent figured out yet, I am really bad at writing devlogs

I kinda forget all that I did in this 4 hours as it was spread around 2 weeks but most it was on the design of it

I have attached a quick sneak peak of one of the animations that would be in the hero section (hope it looks good I spent almost 2-3 hours on it )

Second devlog

So idk how I spent 4 hours already but I have mostly worked on implementing the pipline,enrichers and alot of backend stuff

So far I actually implemented the database schema and defined the types and I would say the end-end pipeline from source to DB is almost done and would just need a bit of fine tuning and model selection

I’m planning on adding some actual sources (which I would prob keep private for now) once the web UI is done

So now that LLMs are good(ish) to find actual serious vulns in software like Copy Fail,Fragnesia,Zcash exploit,etc and now mythos(or fable to be exact) its very obvious that vulns would become much more common

And getting info about the absolutely new vulns is not trivial and are often seen of X,Youtube or other stuff. I decided to build a platform that would be able to find and classify new info from pretty much everywhere on the internet

I first had this idea back when mythos was a first “leaked” and the whole world went crazy with antrophic claims of how its about to “change” cybersecurity

I’ve been planning out the design spec for almost a month now and finally had the time to actually start working on it

And after procrastinating for quite a while I’m finally done with the design!!!

The time spent on this devlogs accounts for the defining what pandora is and all the under the hood stuff, scaffolding the the repo and making sure my agents dont go schizo and generate slop (yes I used AI for alot of stuff 😔 but ig its the future)

Oh I also spent a good amount of time manually making the cool banner and gif! (not tracked)