refactor: move model configurations into models module

[stuckinforloop]

Changes

• Moved LLM provider configuration classes out of summaries.py into dedicated modules:
• Created models/base.py with LLMClientProtocol and LLMError
• Created models/config.py with OpenAIConfig, AnthropicConfig, and GoogleConfig
• Updated imports in summaries.py to reference new locations

Addresses comments from issue #23, specifically the following:

I very much dislike our current summaries flow, which is conditional-heavy, and would benefit from moving the provider configs out. I also don't like, e.g, kit.summaries.AnthropicConfig: these configs really have nothing to do specifically with summaries. So moving them to another place is good

[vercel]

@stuckinforloop is attempting to deploy a commit to the Cased Team on Vercel.

A member of the Team first needs to authorize it.

[stuckinforloop]

@tnm can you share the linter rules. I'll update formatting in the summaries.py accordingly.

[tnm]

@stuckinforloop Thanks for this! We're updating our linter/formatter now; will share the latest shortly & there will be some fixes to the repo. If you can adjust after that, that'll be great.

[stuckinforloop]

@tnm I've rebased the PR against latest main.

[stuckinforloop]

I feel we should also move the get_llm_client func and other LLM related functions to the models module. Keeps LLM related code together.

[tnm]

I feel we should also move the get_llm_client func and other LLM related functions to the models module. Keeps LLM related code together.

Agreed!

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
kit	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	May 13, 2025 7:55pm