Exposing EymenWSMC

NobodyImportant

Notice: This article is in no particular order.

In a series of discord messages in the "Eaglercraft 1.12" discord server, Eymen admitting
to using artificial intelligence, however, he greatly understated the extent of his uses as seen in further topics in this article. The messages have been pasted below.

Image shows a conversation in which eymen admits to using AI for "certain tasks". He also used his ability to use renderdoc as a defense, and then later told radman he didn't know how to use renderdoc (explained later in the article).

As seen in the image, he became slightly defensive when asked about AI despite his name never being directly mentioned. Meanwhile, one of his commits is shown to be comment removal (supposedly generated by AI) in order to cover his tracks. Prior to this, nobody had asked him about his AI usage, and only became defensive once Peyton stated that he was tired of "all of these AI slop ports". Image shown below.

This was part of a larger, 920+ file commit which mostly consisted of only whitespace changes, not meaningful code updates.

An additional image showing the extreme defensiveness when accused of AI

Another commit to his 1.14 repository shows extensive comment removal (for seemingly unmarked things, which were already present in the original fork of the repository. This was most likely the result of an automated mass comment removal instructed by a script or AI) and removal of basic functions. GLX Class got nuked in this commit as well.

Copyright header removal (mass comment deletion)

His port is also filled with seemingly useless no-op classes (a class that contains no functional use).

This is not a no-op class, but rather points out that it contains no-op functions.

The DataFixerUpper class was also nuked, with an obviously AI generated comment, further proving that the 1.14 port is riddled with AI generated code, if not the entire port.

AI generated removal and comment replacement

Useless "optimization" for two functions that get called the same number of times per frame. This is most likely from his sodium port, which is obviously AI.

The entire ports source code is inconsistent in it's inner-workings, and is constantly showing counter-intuitive methods and absolutely useless changes and AI comment removal (there is a lot of this if you haven't been able to tell already). As an example:

Entirely useless check right here. This is probably his way of fixing a TeaVM division bug or fixing a sound manager bug Eagler 1.5 had. That's a band aid fix probably implemented by an AI that has no context of the actual bug, and based on the bug's crash report I would see why it would think to do this.
This doesn't fix the actual bug.

Pretty sure this isn't even needed at all and was probably the AI adding this check everywhere possible because it didn't know exactly where the bugs came from. Because of a TeaVM bug, there should probably be checks before dividing instead of afterwards

This is a game breaking change where he screwed with Entity AI somehow when he (the AI) was making changes to VillagerEntity.java for no apparent reason:

Critical entity data manager bug affecting multiplayer and singleplayer across all ports, with non-vanilla logic and endless band-aid fixes. The bug makes the client and server have mismatched data manager ID's and it's going to be sending and receiving a bunch of incorrect data because of it. All of his ports have this bug.

Sleeping logic being executed where it shouldn't

Another extremely obvious AI generated comment. On another note, the entire fallback system is so obviously AI generated and not vanilla at all. All he had to do was send the assets to the integrated server on startup.

Band aid fixes for incorrectly mapped data manager values

It is also important to note that in other classes of the 1.14 port source code, code is written as the following instead of actually importing the classes required to use it; an obvious AI oversight:

new java.io.OutputStreamWriter(...)
new net.lax1dude.eaglercraft.internal.vfs2.VFile2(...)

More Turkish AI comments are present in vanilla classes that should not have been touched at ALL when doing a port. Image below.

(Also on a side note, he has an extremely fragile ego when it comes to his "ports", which is completely unjustified considering the amount of content that is being discussed in this article today).

Context for this next image: He does not know the changes in his own commit, and lax1dude did not use AI in his original 1.5 or 1.8 ports at all; blatant lie by eymen in order to make his AI uses seem less obviously bad.

Lies and ignorance to his own commits. Also. EXTREMELY important to note: The change was published just over a week ago, not two years ago. Another blatant lie.

In this screenshot, eymen claims he decompiled a modern GUI when it is clearly a recreation, excessively signaling that he did not create it himself (hence the lie).

In the discord server, he brought up how he uses renderdoc to debug his ports, when in a private message, he immediately contradicted himself by stating that he did not know how to use renderdoc at all.

Eymen using type adapters written by Peyton for his 1.17 port (The resources shown in the next screenshot never get compiled to the EPK or read to the desktop runtime, a mistake commonly made by AI or a very inexperienced developer in this niche). When asked, Peyton said these were never made public; aka. no permission was given to use it.

Resource references for the previous attachment.

Further proof he is using code from Peytons 1.12 port instead of writing it himself. Unconfirmed report: He is using 1.12's old update overlay written by ServerDotSo.

Here, He's emulating multiDrawArrays by spamming draw calls. It's a nightmare from a performance standpoint. He (or the AI) wasn't aware that the krono's multi draw extension isn't supported in OpenGL ES 300 so it's ALWAYS going to fallback to the slow, manual emulation. This is a pretty common AI hallucination

The following section is also important to take note of:

Eymen worked on Astra Client v1 until the release of v1.6; after which he was fired for writing the socials backend system with Artificial Intelligence. This can be confirmed by looking at the source leak for the client before the v2 rewrite, which clearly indicates massive usage of artificial intelligence.

As of September 08, 2026, he no longer works on this client and it is unconfirmed his involvement if any in the development of Astra Client 2.

With that being noted, the short amount of time between being confirmed to using AI for everything and the accusations and denials of the messages provided in this article, is not only reasonable, but very likely to assume that he still uses the same practices with AI on his ports and projects he is currently active in.

One of the most obvious red flags is how he managed to spit out entire modern ports, including a full Forge/Mixin port, in just a matter of weeks, which was complete with massive plugin classes stretching several thousand lines long, all while claiming zero AI assistance. Experienced developers like Lax spent nearly two years building a solid, polished Forge port, and took over a year of development on 1.8 and other versions before releasing any beta versions at all. When you factor in that his rushed ports are notoriously buggy and suffer from abysmal performance, the claim of purely human work completely falls apart.

On an unrelated note, he also leaked the source code to 1.12 and flat out denied it when he was confronted with it despite having screenshotted proof, so he should not be trusted to work in any co-developer environment. He then ratted radman out to get him banned. This was most likely a desperate attempt to try and not get himself banned for the leak.

Exposing EymenWSMC

Report Page