Peder Holdgaard Pedersen

> I did think on this one quite hard, but I think mitigating are: > > * Relatively low take-up of Prompts/Sampling Features reduce the risk. Those using the features...

> Most inference APIs (OpenAI's chat completions, Claude's, but also OSS in HF transformers and llama.cpp) require or assume a strict `assistant` / `user` alternance in messages, with message content...

Created PR #173 for this.

I would greatly prefer a way to use tools that is not tied to the assistant concept. Most LLMs today support tools as part of the interface, so I would...

The `IChatClient` in this library isn't fixed to a single model like it is in the OpenAI library. Instead you pass the `ModelId` in the `ChatOptions`. If you don't want...

You could write an additional wrapper around the `AnthropicClient().Messages` and re-implement `GetResponseAsync` and `GetStreamingResponseAsync` to inject the caching. Then when creating the delegate for SK do something like: ``` IChatClient...

I think a good follow-up to this would be to allow `instructions` to be a Resource reference, as having a large chunk of context inside the server info is awkward...

> That’s not an MCP limitation is it? (roll your own MCP and chat interface and there’s no issue rendering video or html or whatever), that is a UI of...

Regarding `ui/message` and > Host SHOULD add the message to the conversation thread, preserving the specified role. To me this seems like something that SHOULD require a user approval. Giving...

It would be very useful to have a way to push context to the host application without necessarily triggering a user message. Something like `ui/notifications/context-update` that the host accumulates and...