Feature Requests

For thinking models, the contents of <code> are useful but do not necessarily need to be shown always. Hiding it by default and showing it when clicking the button will greatly improve the readability of the response.

I love using the TAB shortcut to change model or CMD + P for my presets. But most of the time I'm pressing too fast and I miss the model or preset I want. I would love to be able to add SHIFT as the shortcut modifier to browse the model (SHIFT + TAB) or the presets (SHIFT + CMD + P) in the opposite direction. Feel free to tell me if you need more details or if my message is not clear enought. Have a great day !

Just a shortcut that will switch between the context or start blank rather than having to click.

I know this is a big lift but it would be awesome if somehow Kerlig was my main MCP client with an option to choose the inference server like ollama or OpenAI

I hope this email finds you well. I'm reaching out to inquire about the possibility of integrating the latest Anthropic model, Sonnet 3.7, into Kerlig.

Current design: The pin feature forces the interface to stay on top, which covers all other software and prevents me from copying and pasting between multiple applications. I have to move the Kerlig window over, which is quite cumbersome. Note: The interface UI does not disappear, and pinning the window to stay on top is clearly two separate feature requests. Since we are afraid of the interface disappearing, we choose to pin the window, but once pinned, it always covers other applications. Strongly request that the pin function be modified to "not disappear the interface." Please have users who are concerned about this issue vote together, as the developer's updates are extremely slow. Most requests in this community have hardly been implemented by the developers, which is regrettable.

A common workflow that I have is take a screenshot of a slide on Zoom, then ask the LLM to "Extract the text so I can take notes. If the items are in a list, put each item on its own line and don't use numbered or bullet lists. Don't add any of your own commentary." I'd like to avoid writing that each time.

Hello, it would be great for Mistral AI users to have official access. The official API is really very fast in Europe compared to other providers.

A useful feature would be the ability to specify a generic, OpenAI-compatible API in the settings. This could include automatically fetching available models or allowing users to manually select them. This way, other OpenAI-compatible services can easily be integrated.

The idea here is just to support any Image Generation model. Stable Diffusion would be a good start.