This mod has bugs with current alpha where it breaks the continue main quest button now that it is divided from quick choices.

any plans on updating for current alpha or just waiting until its fully out?

Yes, works with free. The realism option adds a lot of tokens, and I haven't tested it lately so it might actually work with free (because I reduced token usage by 50% (even less than vanilla version for summaries) - but I am not sure. You would have to try it with free and see how it goes. If it doesnt work with free you'll get a red colored message saying tokens were cut.

Is this work with free models ? There is writing sapphire next to realism option.

thanks @Totes :)

Just released 1.51.0

- initial support for Alpha 7 (new grid system is interesting)
- tightened up suffixes (you dont get these automatically, you'll need to go to options to reselect your suffix if you were using them)
- small tweaks to event checking
- updated linear suffix to work as intended. (fixes to deviating from task)

Most important mod anyone could get for this game,

Also, super awesome, helpful, and very patient dev :3

Proud to present 1.50.0 - which includes token reductions, even using less tokens than the vanilla base game. Description optimizations for entities, as well as on state updates. Improved compatibility with other quest related mods. (Less reliant on subtasks so other quest mods that over-write files are not negatively impacted) Optimized for other languages as well.

Thank you :)

1.49.1 coming soon with additional token reductions (50% cuts in some areas) - still in testing but so far I'm seeing no loss in game quality, just faster text completions since the token overhead is lower.

Ah yeah. So what happened was 1.48 removed subtasks in favor of multiple tasks to achieve that end, but it actually ended up causing a lot more issues, so I had to revert back to 1.47. However, 1.48 also had some other improvements (which I will update in the description shortly as well) that were lost, so 1.49 was merged 1.48 changes without anything related to the subtask removals. Most of the changes were reduction in tokens and minor tweaks to system preamble, faction generation, and quest openers and rewards. this would have been a patch update (1.48.x) but since 1.48 added a new feature that no longer exists I opted for 1.49

Sorry, but I think you didn't mention what was improved in 1.49 in the description

Another massive update just launched! I've spent a whole week on this one with extensive testing. 1.42 overhauls the quest system yet again. Tasks can now be up to 9 steps in length, the story is much more cohesive, and the opening is much stronger when starting with quests enabled. End goal quests do not have options at the end, while open quests do. Event checking enhanced, location transitions and object permeance are tracked better. A new reactivity suffix was added for linear mode, which focuses on the quest more without interruptions. You should now have a much clearer understanding of WHY things are happening or WHY a task is present. and tasks can easily be rejected if you want to change them. THis change might introduce more "sudden problems" - so using a Reactivity suffix is advised if you dont like that behavior, though safe areas should remain safe. If any issues, let me know and I will fix it.

1.41-2 massive cut on tokens, and also removed some prompt overrides for event checks since the new version handles those a lot better. This also reverts back to most of vanilla for entity generation, and only modifies description of entities to be more rich for image generation and also provide immersive, perspective based descriptions. Quest progression is still experimental atm, but it's been difficult to find a sweetspot. Tasks can still drift because not allowing them prevents a lot of features in the game from happening. However, Tasks will fail if drifting enough, and will make further tweaks.

1.40 re-introduced more tokens - only because 1.39 was too easy and every quest was just almost automatically compelting and combat would just resolve quickly in the player's favor. This time each quest task feels much more meaningful, taking more steps and don't get marked completed until near the end of task completion. Tested on several playthroughs and each worked very well. Seems to be the sweet spot but more testing and feedback is required to tune it.

1.39 features a much better quest progression system with better vessel support (more details in mod description)
1.39.1-2 fixed some character generation instructions, emphasizing on what the player can see and also optimized for image generation AI
Finally 1.39.3 cut back on token usage.
Would love feedback, as the token usage cut might have had some unforseen consequences. Particularly around the quest system. So far it's been working great for me but I'm only a sample of one.

The combat works fine now. I think I got some really weird (bad)luck earlier. Or probably the base game is being strange. I don't know how much a mod can overwrite things.

I tested 10 fights already, and no weird stuff seen.

if its multiplier or numbers related, check out the default/misc folder, there seems to be some txt files like dmg-mults.txt and dmg-rng.txt you might be able to tweak. not 100% if they are used but I suspect they are.

Ah, yeah that might be related to the multipliers used (this mod doenst touch them) - if you are taking damage but the amount is low, then its a multiplier issue. If you are somehow just always dodging or the plot always moves in your favor, then it might be the instructions in the mod.

The difficulty; Adventure, Roguelite and Insane; I think it's a bit scuffed.
Insane I get so much damage, but on Roguelite and Adventure, I rarely if at all I get damaged. ?
So might not be a problem of this mod.

Playing on the dungeon crawl mode, not the immersive.

1.38 released - I believe I fixed the nerfed combat issue but its not fully tested. Instructions that prevented the story from moving beyond the conclusion of a task had a side effect of preventing negative consequences (even taking damage) - Progress Alignment now works more like 1.36 - which attempts to avoid pushing the story past a task result but does not over-correct to keep the story there. Combat should work more like before now, but story can drift. You can increase the challenge as well by going into options and selecting immersion or hardcore reactivity suffix.. Still more testing is required and will be pushing additional updates as issues are found.

Again, love your mod.

Thanks for the feedback! I overlooked combat while being hyper focused on story. I will see if I Can resolve it and if not roll back the last change. Appreciate the feedback

I have tested the combat a lot. Doing about 100 fights. I seem to get little to no damage.
I do damage during my turn, and during the enemy turn also sometimes.

The base game is a bit over the top, but with this mod, fighting is way too easy.

1.37.0 features massive improvements to the task system, where the story stays relevent to the quest unless you are not tracking it or make deliberate actions against it. Options make a lot more sense and deviations are subtle. This greatly reduces story inconsistency issues and makes the story a lot more comprehensable. Also you still have 100% freedom to move away from it if you choose, and it dynamically alters the story via updating quest tasks more intelligently.

@SZ This does not touch plausibility so you should be good to go

Does this effect plausibility, only asking because there is another mod that affects it, so just wanting to see if they were compatible.

@Reffy thank you so much!

I want to mention that this mod fixed a huge problem of the base game.
Without mods, the AI constantly generates new threats. I can't even talk to an npc for 1 minute, before "evil shadows erupt from the ground."
It doesn't matter how many enemies and threats you defeat, a new one immediately pops up. Making you fight forever.

Very good mod. Should be essential.

as of v1.35.0 I am proud to present the new progress alignment safeguard system with a ton of improvements to quest and task generation, along with subtle changes across multiple prompts that ensures quest/task tracking stay in sync so the story never advances past your current progress by accident. This feature is only used when you are actively working a quest, and still allows full autonomy. It prevents the story from moving beyond the closure of the current task, unless the task was rendered impossible or failed. It took me a ton of hours and testing to get right, but works perfectly in A6.6 with sapphire. I hope you all enjoy it and let me know if there are any issues!

oh yeah it is read the previous comments

is this working on the alpha?

As of v1.32.0, it is recommended not to use any suffix at all with this mod if you are using Sapphire tier. For free tier, you should still try it without a suffix, but if you run into issues with npcs never talking, you can try the "Chatty" one.
Really the model has improved so much since this was first introduced, I may deprecate this feature some time in the future.

Also improved task suggestions to be less spoilery and task checking to be a little more strict. Feedback is appreciated.

As of v1.31.0, MBTI has been descripters have been removed. While I like the idea of using those on the character description cards so the AI has more to work with, it does so at the cost of muddying the story telling and ruining immersion. I dont particularly like knowing someones personality if I just met them. Additionally, there were too many times they weren't honored as the story itself takes a much higher precedence. Also, Master Actio created an awesome mod that does a lot of heavy lifting with MBTI which I would recommend for those that gotta have it. It should work with this mod as well as long as it is loaded after but I havent fully tested it. you can check it out here: https://psteamcommunity.yuanyoumao.com/sharedfiles/filedetails/?id=3525261594

v1.29.0 released and is running well (even a lot better) with the current Alpha (6.6 as of now)

As of 1.28.0 - its advised to reset the action suffixes as the preamble has been adjusted to account for advancements from the AI. One key issue was instructions to add some variance would cause some of the story to become disjointed as it scrambles to follow the instructions along with tell the story. This wasn't much of a problem before but after the latest updates to the AI model (sapphire) it became more of an issue. You can still get the variance by using the instruction suffixes, but tests without using any (just default) have proven to be much better at least with sapphire. The old behavior still exists when using suffixes.

I've updated 1.25.0 and 1.25.1 to use the new prompt templates for the latest version, restoring descriptions. I"ve also tweaked event checks and have improved npc tracking slightly (npc following you directly or moved to another area with you) - not full proof but much better. When debugging I noticed that often times complex names were being truncated or summarized, which then caused the game system to not be able to interpret that someone follwowed you. Ensuring more accuracy around using "Existing Characters" list properly.

Working on an update, and right now this mod doesn't change any prompts around checking location changes anymore. Still going to see if I can do anything

I've noticed this as well, it seems like it is hit or miss for me. sometimes it moves me and sometimes it doesn't. Given the game has changed a lot since my last update - I am considering removing the custom checks I have for checking location change.

during the quest, it doesn't seem like it's moving me to new places anymore. Like, the quest in some place, but I never moved from the initial point .

Im playing the game since the begining of early accers, this las patch is a game changer, the coherence and creativity has reached a new level, Im really enjoing it now. Very good work

Thanks for the reply. Sometimes the AI doesn't even both updating the state of the NPCs. They'll just keep talking and acting like they are there without every moving the NPC to the new area. To some degree it almost seems like the AI doesn't know that I've changed locations.

Wanted to add: that the 'bad judgement' issue is something I've tried to mitigate as much as possible, and the following thing happens in vanilla as well. This is why the game comes with settings that disable the game changing the state unless the player confirms it, because it's unreliable how it works. All state based effects are done by asking the AI in a batch of numbered questions and only contains the most recent passage of the story along with some current state data, it does not contain any summary of the story and the system prompt is much different. So even if you use a custom model it is still hamstrung by the developers design decision. However, it probably would be very expensive and slow to update all the game state using a higher reasoning model along with a full summary of the story (which is reserved only for the story teller)

@Harkmagic, the AI determines who is following you after reach turn when it is asked the to list the characters that should be following. based on the previous passage, if the AI thinks they should be following you, then it adds them to the list. There aren't any deliberate instructions telling it that npc characters are following the player, but it is possible the AI makes a bad judgement call. Especially becuase a cheaper model is used for "status checking", which causes bad judgement with questions like "did you take damage?" or "did you make significant progress on the current task?" - the model for answering these questions is different than storyteller and less smart, and also has no context other than previous passage and state.

Is this mod what is causing every NPC I encounter to become a stalker? Is an NPC gets mentioned and I move to another area the AI will either interpret the new area as being very close, effectively walking across the room, or the NPCs will get up and follow me to the new area. Either way the effect is the same, if an NPC gets mentioned they will follow me everywhere.

It does work with local text models, I use Kobold with mistral small 24b instruct 2506 ud Q6 k xl. with KoboldCPP.

With that said, Sapphire is phenomenal and it works best with it.

so i just came to give you a good laugh i subscribed and proceeded to spend and hour figuring out why this wasnt working.....i didnt turn it on. i literally will spend days modding games and i somehow forgot to activate the mod. i have made mods for my own personal use in other games just by looking at other mods and using similar methods to make it the way me and my brothers like and i still forgot to turn the mod on. i was so annoyed at myself that i went and took a nap.

It works, but it slows the generation, and make it require a lot more VRAM, but i can confirm it works, and also from my testing to work proprerly it needs more Context size (or tokens) than normal, but it is so great, and feels much better the generation.

I've only tested it with AI Roguelite AI's and ChatGPT 3.5. But would be curious to find out, if you don't mind testing it out I'd be interested to learn about your findings!

Does it works with local text model? Like kobold? Or is just the online APIs?

Forgot to add, you will know if you exceed the token limit or not as the game will output a message stating it. So if you try the reactivity realism one and you don't get any messages stating the tokens were truncated or shortened, then you should be good to go.