Thinker is an AI improvement mod for Alien Crossfire I've been developing for a while. Some of the features include:

* New production AI for bases
* Improved terraforming AI (lots of condenser-farms, boreholes and forests)
* AI builds more crawlers and deploys them better
* AI will deploy colony pods much sooner instead of wandering on the map
* Customizable options for free formers and satellites for AI factions
* Does not affect player-run factions or game mechanics

Project homepage and latest downloads are here: https://github.com/induktio/thinker

You can also email me any comments. Address is in the release notes.

Thank you! Looking forward to trying this.

Sure, you're welcome. Just to note, this mod is still in active development. I consider the code to be pretty stable currently, so it's mainly a question of which features to add or rebalance in the future versions. Also terraforming and base production priorities might receive some tweaks.

Generally speaking the mod shouldn't include things that change the game mechanics, like modding the tech tree or changing the combat values, but any ideas on how to improve the AI behaviour are appreciated.

It's always great to see someone working on the AI of this great game :) Some things which come to my mind immediately:

1) especially the AIs without inherent extra unit support (or those unlikely to get it from policy choices) often tend to cripple themselves productionwise, because the units are literally eating up the last produced mineral. The AI should stop unit production then or rebase units to other bases if possible.

2) Morgan AI suffers both from 1) and often even more because it forgets to expand - which is kind of a death sentence, if a special rule prevents bases growing past 4 citizen without hab complexes

3) AI doesn't know about how to make good (sea)formers - they never seem to add any armor to them. Maybe it's just me, but I think formers with inbuild 3-res-armor are a good investment vs. the mindworm/i.o.d. threat

4) AIs don't cope well with production switches - I see them often waste a lot of production by switching (you can provoke that by killing garrisons during war time)

5) I have read about this project not intending to include balance changes; however then for my taste the AI needs to learn about the big power of some game elements. The already implemented better use of advanced terraforming improvements is one thing, but as important ist teaching them the power of forests from the midgame on as well as the good back-for-buck relation for any chopper-chasis units.

And finally a (probably stupid) question...would it be possible to somehow include the content of the existing SMAC/X AI patches or to make this compatible with them? I'm aware that this is probably not possible or at least not something easily done, but I see a situation coming up, where one has to decide between having your advanced AI improvements in the game or the existing patched, whch tackle quite a few bugs and balance fixes (beisde improving the AI as well)

Some of these points refer to issues that are already solved in v0.6 release. Let's address them in more detail:

1) Running out of minerals has been maybe the biggest problem of the AI, and this patch should fix it for most practical purposes. The AI simply stops building military units if it doesn't have a certain mineral reserve available.

2) Morgan AI expands roughly equally to other factions in this mod. Generally the biggest deciding factor on how big the AI will grow is the size of their starting location landmass. Next version will bring a new feature to work around this limit, too.

3) It doesn't currently add any armor on the formers but it can compensate with the numbers by building many times more formers than the standard AIs would have.

4) Production switches happen now usually only during secret project cascades.

5) AI should start forest planting immediately now.

About the other patches, there are some bugfix patches for SMACX but none of them really seems to improve the AI in any noticeable way. Scient's patch does bring some bugfixes so I'm considering merging them later, but the problem with binary patching is the maintainability of the code.