Prasad Chalasani

Agree with the original post -- I'm at a loss on how to create a tool `foo` for my open-source library `blah` so that people can directly run `foo` using...

Thanks @zanieb I somehow didn't find this page in the docs, but found the same instructions elsewhere. Yes my library `langroid` is published, and I have a separate examples repo...

> This is now resolved in the latest Claude Code (`v1.0.35`) and plugin (`0.1.11-beta`) versions. I guess we still need to migrate the various JSONL and other files from the...

Thanks for the interest. Is this an LLM gateway sort of like LiteLLM, langdb, portkey etc?

@OctavianTheI can you give a short outline of what such an integration might look like? I'd imagine it would be similar to openrouter, langdb or portkey.

Ok, an integration similar to openrouter would be fine. Besides the code change, we'll need a docs page and some basic tests (in test_llm.py)

We've already starting using `MockLM` in some tests, we should carefully look at all tests and see if we really need an actual LLM.

@coretado the only general thought I have is -- are the OpenAI and Anthropic really so very different that we can't factor some commonalities and avoid near-duplication? I think you...

> Is your preference more for `openai_gpt.py` housing additional functionality to handle Anthropic specific code paths, or to lift common functionality and call where appropriate for two classes representing OpenAI...

> * should look to add Anthropic focused tests to? Yes > * for both chat and completion; are we looking to primarily support Claude 3.x onwards? Don’t worry about...