TODO

- new branch: feature/function-calling
- add support for image inputs (path or paste from clipboard)
- add support for function calling (dangerous) (the command will be executed represented by the user)
   per msg function calling allowance, request approval from the user
  -- bash / zsh 


- new branch: feature/responses-api
- switch from Chat Completions to the Responses API
- add Tools supported by this model when using the Responses API.
  -- web_search (by openai)
  -- Image generation (by openai)
  -- File search (by openai) (require uploading file to vector database)
  -- Code interpreter (by openai)
  -- MCPs
- add config options for:
  -- models
  -- max output tokens
  -- allowed function calling
  -- allowed tools
  -- config MCPs