I thought about that, but there are a number of problems here:
1) Custom defences: if you don't have the same mod set as blueprint author does, you probably won't get all defences working and it won't require significant efforts from you to take over the base
2) Landscape: the mod does not save landscape. When you cut a few rooms it does not matter. When you spawning a whole base and it have one completely undefended side because there originally was an ocean or a mountain, it's also an easy to take over. Turrets can be spawned inside a mountain, etc. Storing and using original landscape is also not always good, because you have to consider roads, rivers, shores, mountain amount, biome, etc. There are plenty of maps in the database, but not so much to provide a variety of bases for each possible combination of all those tile properties.
3) Base quality: the mod can't determine actual base quality. The blueprint can be either well-defended real player's base, but it also can be a dump of someone exploring the ruins, and this is much less interesting.
4) Too high value. Most well and interestingly defended real player bases have way too much loot inside. Balancing this leads to either less interesting base or to enormous defence forces. Either way seems bad for me.

So actually this is top four blockers of this feature.
Possibly there is a way to look through all blueprints I have and select several dozens of really well bases by hand and use them as outposts, but this will sovle just a part of all possible problems. However, if you have any ideas on that, please share. Maybe we can come up with a solution to those problems.

Hmm... I can see now how this may take a lot of work to implement. Talking hypothetically, this is what I have so far in terms of possible solutions to these problems.

1. Perhaps there's a way to streamline this? Replace unknown assets that are used in a mod with a generic vanilla weapon, and the same with mortars and turrets. Of course, for mods that offer totally alternative methods of defense, the problem may not be as clear to solve.

2. This one I think is the hardest to solve mechanically. A possible but not necessarily practical solution is to just encourage the submission of more maps to create a better variety of bases to fit certain circumstances. Perhaps if there are minute differences in a certain tile and a certain submitted base these could be disregarded? I feel like there is a solution to this somehow, but it may require a lot of work.

3&4. Maybe for these, it would be possible for the player to set minimums/maximums of the value of bases they'd want to come across, and perhaps with some room for exceptions as well? I feel like the problem doesn't necessarily lie in the fact that boring maps and contarily very hard to take on maps would exist, I think it's that one would be in an abundance. Plundering a variety of bases only to at one point find a sprawling empire could be a really cool feature, but I think this relies on the balance between the values and qualities of bases. Problem is, how can there be a system to efficiently and accurately balance them?

I don't know, this presents a lot of problems, but I feel like if we could find solutions to them the mod made as a result of it would be really amazing. Honestly, I also like the idea of hand-picking bases, obviously it requires a lot of work as well, but it is a simple solution and with enough dedicated people might actually work out. Personally, I'd be glad to help out with a process like this, but it is up to you how you want to implement the feature.

I'm still thinking about it, but still don't have a solid enough vision to start doing something.
The first problem is the most simple and, Steve, you're right, it can be replaced automatically based on item title even if there is no corresponding mod.

Cost estimation also has some complexities: calculating cost before placing the blueprint is simple and reliable, but slow. It can take up to a second to process a blueprint, and if it doesn't fit by cost this time is wasted and you need to pick another one. Picking 50 blueprints in a row is almost a minute which is too much. Precalculating cost before upload is fast, but not reliable: if you don't have glittertech mod, alpha-poly walls become wooden and 10^8 base worth will become 10000, while the mod still thinks it's 4 orders of magnitude higher. This problem could be probably resolved by expanding server to something more than just a storage bucket, but this is way more complex.

And, anyway, it doesn't resolve the most difficult item number 2: landscape.
There was an idea somewhere on reddit to use ruins from the same planet. I.e. when you have the same seed and the same generation size, you'll get the same planet layout. So if you find a blueprint with the same planet seed, you can definitely place it to the same cell and it will fit perfectly (if there are no another terrain generation mods). I started to gather planet seed hash to check if there are at least two maps with the same seed to check feasibility of this approach. But I really doubt there would be even two same hashes.

So, I'm going to think a bit more and gather more info to make next steps. All your ideas are appreciated!

Perhaps you could provide a seed, or small set of seeds, that players need to use in order to use the feature.

A very hypothetical solution to the issue of bases being overpowered in terms of the material of the base (seeing some huge base made ENTIRELY out of gold) would be to maybe write a code to detect the base material and if there is too big of an only gold base then just exclude it from the database of saved bases.

I dont know if this is already implemented or not but maybe you could start storing the bases previous wealth factor and using that determine the bases quality. You could also record the size and material the base is constructed out of (again the gold thing) and scale bases with a quality scale. Maybe if your own bases wealth level is pretty high then you start to find wealthier bases, but if you had just started out lay back on the turrets and give people wooden bases with maybe a bed or two.

Also you could delete random items from the stockpiles to balance it out aswell. On the contrary you could make it detect what is in the stockpile and use that to determine what quality the base is and again restrict that base from early game players.

I can imagine this being incredibly difficult to code and I thank you for making this wonderful mod already, and I hopy you take my ideas into consideration

The scaling concept is a really good idea. I think what really makes this a hard issue beyond coding the features in, is how the player interacts with them. For something like this, it's a question of, how much do you leave in the player's control? I'd imagine its hard to draw the line between giving the player freedom of how they play the game and what is reasonably codable into the game.

was about to ask about this feature and found this thread on the faq... sad that its not happening :( ai bases are just boring squares

Originally posted by An Ol' Spicy Keychain:

was about to ask about this feature and found this thread on the faq... sad that its not happening :( ai bases are just boring squares

I won't change vanilla bases because I can't guarantee that each ruin is better and more interesting than a vanilla base. However, you can check planetary ruins feature which adds basically the same, without removing vanilla bases. You can't trade with new types of factions' settlements, but in all other aspects they are pretty much the same.

Originally posted by MR. TETEREW:

I won't change vanilla bases because I can't guarantee that each ruin is better and more interesting than a vanilla base.

"Each" and "guarantee", no. But considering how boring the vanilla bases are, I think you can assume that most would. I find this an intriguing idea worth a couple of trade offs.

Guesstimating value (upper/lower cut off) and replacing defenses seems a reasonable approach. Maybe maintain a translation table for the most common modded materials and defenses, so alpha poly can become plasteel instead of wood.

Terrain is the real kicker. I wonder, would it be possible create a local map from a different seed? I.e. you save the seed (maybe coordinates, or whatever else is used in map generation), biome, mountain type, road/river type and orientation, and adjacent water tiles. Then fetch a matching base and generate the map with its original seed. On paper at least, that would solve the placement problem.

It would really add a lot to the game.

Hmmm...

I love-hate this idea. AI bases need to be designed for AI. So bases put forward should be put forward intended for AI use. Rather than collected from random playthroughs.

If cherry picking map files for use as AI settlements were ever deemed viable, I would happily volunteer my time to that project.