Typing Mind

Sorting chats resets back to "Recent Chats First" every time Typing Mind is refreshed. My request is that the sorting setting should be permanent (i.e. if I sort by Title A to Z, that sort is saved and permanent until I change it back to another sort), and not reset back to "Recent Chats First". As an alternative, add a keyboard shortcut option to match a sorting option to be able to apply a sort through keyboard shortcut.

Two feature requests: It is very inconvenient to have to manually to drag scroll bar to reach the latest chat message without option to jump to bottom of chat using button like ChatGPT. Similarly, it is inefficient and inconvenient to scroll through entire user message text. Suggestion to truncate messages to a reasonable length with option to expand, similar to ChatGPT and other major platforms.

We can have option under each results to chose which results to reinject in discussion with LLM (to save tokens) By default all previous results are send back to LLM, but the more we interact, we don’t want somes to pollute the discussion… I know we already have option to send n previous messages, this one is more granular

I often want to ask clarification or dive deeper on part of a response. I would like to be able to create a quick side/sub-thread - without losing the main thread. I for sure don't want to send the whole context (which is often 100K+) just the response I'm asking a specific follow-up on.

This allows you to view multiple chats with different LLMs. This can be great for comparing outpus of models as well. Also this allows you to quickly use the LLM for another topic that you don't want to add to the current chat.

Can we have conversations where we can choose multiple bots to talk to - similar to Silly Tavern? https://sillytavernai.com/ For example, then we could have brainstorming sessions with different bots with different skills, or learn languages in a room with multiple bots for more immersion. This will also open up the program to more people.

I hope to use a cheap model to generate titles. If I use the o1 model, then generating titles with the o1 model will be unaffordable

Hello , is it possible to define a model for title and auto sugest for not use the best model for that . for exemple , if i use o1 , i don't want paid an o1 request for this but use gemini flash or other with ope,router

Give ability to agent to use Typingmind commands inside chat, such as moving the conversation to folder or add a tag Example of usage with queuing system: The user ask for a recipe of a chocolate cake and move the conversation to the right folder. Your are a chef, give me the recipe of a chocolate cake. ---- Translate the recipe in French. ---- /typingmind-move-to-folder /recipes/cakes Example of commands that could be used and even automated: /typingmind-add-tag {tag} /typingmind-move-to-folder /recipes/ Also, with a prompt, we can also imagine that the agent could be able to automatically determine the folder, add the tag and execute the /typingmind-{command} itself. The previous example then become: The user ask for a recipe of a chocolate cake and ask the agent to move the conversation to the right folder. Your are a chef, give me the recipe of a chocolate cake. ---- Translate the recipe in French. ---- Please determine the right folder to store this conversation. If it doesn't exist, create one. To move the conversation to the right folder, use the command `/typingmind-move-to-folder /folder/subfolder/your-folder` where `folder/subfolder/subsubfolder` can be nested as your want. Please note the current folder are: /recipes/cakes /recipes/pizzas /business/ideation Only respond with the `/typingmind-move-to-folder ...` command

Being able to fork chat with options inside {}, Example of usage, in combination with queuing and agent feature: Give me the recipe of a {chocolate cake | banana bread | pizza | pasta | omelette}. --- Translate this recipe in French. --- @[TO_JSON] { "recipe": "...", "translation": "..." } Here, the chat would be forked into 5 differents chats, one for each recipe, and then for each chats a queue would be created to process the request.