Leidtier

In the HTTP branch, the sentences are produced async and put into a queue. Every request to continue from the game simply returns the next sentence from that queue. Interrupting...

Maybe move this to `v0.11.2`? This already confused some users.

I wasn't able to reproduce but I made an attempt to fix this here: #334 I assume it happens [here](https://github.com/art-from-the-machine/Mantella/blob/0fc19dcc12c5115ac557ed076159beabf21c503f/src/tts/xvasynth.py#L206-L213) if the batch synth fails before that for what ever...

Looks like your tiktoken registry doesn't know the encoding `cl100k_base` for *gpt-3.5-turbo*. Similar to this issue over here https://github.com/openai/tiktoken/issues/80. As you are the very first person with this problem I...

@danniesim thanks for the report and the investigation. We had a couple of people installing MGO lately and they didn't report a problem like this... After you ran `Mantella.exe` outside...

Found the issue with the inventory mentioned above. I tested it on an existing save and the way we register for action events on the side of MantellaSpell right now...