A lot of aspects are just confusing / obtuse.

[Dark-Thoughts]

Marinara Engine is supposedly meant to "just work", but I feel a lot of the systems are so badly explained & tucked away, or even contradicting themselves, that they become more confusing than anything else. One aspect are the agents & function calls, and the "game" chat mode.

In particular if I add the Game Master agent to an RP chat:

• Do I then still have to edit / create a Preset that includes the Game Master (Agent) Section?
• Do I still have to enable tool use & add function calls for dice rolls & game state updates? Because they're also already enabled for the Game Master agent itself and I can't tell if that's just redundant, still required, or even making it worse if I "double / triple dip" here.

Same thing goes for "Game" chats. When you create one, seemingly it already does a lot of those functions from various agents, but in the chat settings, none of them are actually enabled...

• Should I enable them?
• Does enabling them there do anything? Does it make things worse?

A lot of the time it seems like agents just bloat context size to the point where things just fall apart. Agents error out and need a re-run and generation often ends up in a loop ad infinitum, probably because it runs out of context and gets confused.

[GhostGuy9]

Using Kimi, it works as I thought it would with all the agents enabled. Since this is still new I highly expect GM working in a RP chat to not work and act like a person over a GM. But since this is still new many functions seem open enough for experimentation.

Using the standard setting and nothing more then my lorebooks, it works so well on what is in one world vs another (Multiverse Style Lore) and how it made the perception filter widget from the lore is amazing. I did notice that it took a long while to generate everything but I have like 10 worlds so i imagine everyone else would have a better time then me lol.

Good luck and honestly just experiment, I assume that what most people do here when new RP engines like this one get created.

[Dark-Thoughts]

This frustration already comes specifically after several days of experimenting. Lorebooks just don't work at all for me, unless I explicitly link them to a character - but unlike SillyTavern I can only link them to a single one, not multiple, nor can I globally activate them.

[SpicyMarinara]

There is no Game Master agent unless you added a custom one named that way, or unless one of my devs added it in the recent PR, but then I have no clue how it's intended to work. You are not supposed to have a Game mode in Roleplay mode. In Game mode, you can only enable a limited number of agents, and they're all described, and the explanation is right there:

As for agents, obviously, you're not meant to add them all, but no one is going to stop you from doing so, anyway. It may be that your model isn't handling them well.

It sounds like you may not be using the most recent version. We introduced many fixes, especially addressing lorebooks. Keep in mind the frontend is in the early alpha stage, and there are a handful of people actively working on it. If you're experiencing any bugs, you should report them; otherwise, we are not aware of them and can't fix them.

[Dark-Thoughts]

There is no Game Master agent unless you added a custom one named that way, or unless one of my devs added it in the recent PR, but then I have no clue how it's intended to work.

Scroll down to Misc Agents. It's the second to last one of them. I've not added any custom agents yet, as I already feel confused enough about the existing ones to feel comfortable doing that.

It sounds like you may not be using the most recent version.

I installed via Git and the shell script is supposedly handling git pulls automatically. The Home screen shows v1.5.6 and the Advanced settings tab also Build bcbfd7d. As far as the release page goes this should be the latest version, at least for the main / stable branch.

If you're experiencing any bugs, you should report them; otherwise, we are not aware of them and can't fix them.

This can be fairly discouraging when being dismissed.
#256

Same for "closing" discussion / feedback threads.

[SpicyMarinara]

These are artifacts; they do nothing. Removed them in the newest version. I figured I had removed them from my local version. The thread you linked is old and outdated by now. The model will work perfectly fine if you select None as persona, you will only be referred to as the user. At the time, I provided the fastest fix for your issue. As for this feedback thread, the feedback concerns artifacts that do nothing and shouldn't have been there in the first place. There's no point in keeping it open. If you'd like to address different topics, you're always welcome to create new discussions, issues, or threads on Discord. It's easier for us to address issues that way than to search through a single giant thread that's constantly updated, touches upon everything, and is hard to track. It's a matter of both convenience and streamlining the process. In other words, this format helps us a lot, and we're thankful to those who use it.

[cha1latte]

‧₊˚❀༉‧₊˚. hi hi friend! i pulled out some of the stuff that still needs action because there is some good stuff here :3 here is what I am noting that I can convert into PRs and take care of for you. ‧₊˚❀༉‧₊˚.

1. Audit the UI for deprecated artifacts and tag them as such. the "Game Master" no-op agent in the Misc list is gone 🎉to do some additional cleanup, lets audit and tag anything mid-deprecation with a "(deprecated)" suffix or tooltip

2. Add a line at the top of Game chat settings explaining what's on by default. something like "Game mode includes dice rolls, state tracking, [etc] by default — agents below are optional add-ons." this will clear up the "wait, nothing is toggled on, is my chat broken?" reaction new users get when the mode is quietly doing work under the hood. a mirror line on Roleplay mode ("game-style features aren't intended for this mode") will help the reverse confusion too

3. Add a sentence to each agent description clarifying preset overlap, and/or surface a soft warning when both are configured for the same behavior. best guess from experimentation: agent + matching preset section is mostly redundant + slightly harmful (extra tokens, sometimes conflicting instructions). a one-liner in the description or a soft warning at config time will let users confidently pick one place for each behavior instead of double-dipping just in case

4. Add an in-UI signal for cumulative agent loadout cost. even a rough estimated-token-overhead readout next to the agent toggles will let people self-correct before stacking themselves into context-window failures and surface that the issue is loadout weight

@Dark-Thoughts could you please check the lorebook scope issue on latest version and re-file if it's still happening? it might be fixed, but if it's not for you, could you please repost that specific issue with repro steps (what lorebook config, character setup, expected vs actual) and I can get it taken care of for you.

have a happy day °❀⋆.ೃ࿔:･°❀⋆.ೃ࿔:･