It reminds me the Death Hand of the Harkonnen... 

Hey @ffm2 - Awesome work!  I made a mod of your mod.  I thought the communication in your version was a little too verbose -- too much detail, and I wanted to quickly answer the question for the host:  "WHOM SHOULD I SWAP" to achieve balance.  I'm attaching the two files which I modified: gui/gamesetup/Pages/GameSetupPage/Panels/Buttons/BalanceTeamsButton.js gui/common/network~LocalRatings.js Mostly, I wanted to add the /suggest command, but I ended adding or modifying four: /evaluate /suggest /suggestteams /about In detail: /evaluate     Shows player rankings, including observers, and evaluates the current two-team setup. It does not propose new teams. Example   /suggest     For 3v3 and 4v4 games, evaluates eligible one-player swaps between the stronger and weaker teams and displays up to two swaps that most improve the predicted balance. Example Output /suggestteams     Shows the current team setup and, in a 4v4 game, several fixed rank-based full-team configurations. It does not show rankings.     /about   Prints attribution and the command summary.   What do you guys think? ps - In this version, I also manually nerf my own LR rating, because it is a little ridiculous.  I might be accused of using this to try to get better teammates, but I'm really just trying to make it as accurate as possible. pps - I'm ignoring edge cases.  The mod only suggests for 3v3 or 4v4 game.   I think that covers 99% of use cases.   BalanceTeamsButton.js network~LocalRatings.js

It turns out it has such a long history

Special attacks with bombers and nuclear missiles: (slowness is from the capture on the notebook, FPS in the top corner of the video). Gravação de tela de 2026-09-26 01-42-06.webm

@AsherIf you select only one map at a time then only that map is converted. When this is however a map pack, the entire pack is converted map by map but into the same folder, same as its source. Sorry, could not help it - just made another release 1.8.0 that provides better entity conversion support including automatic suggestion of most likely replacement and a few other improvements (e.g. @vladislavbelov's tree randomization script). One new architectural thing is that I extracted a common code package that will be also used by other ATLAS-related tools that I am preparing.   Please find the latest release 1.8.0 in the top message of this thread (with updated screenshots)