Add multi-provider LLM support (OpenAI, Anthropic, Gemini, Databricks, Ollama)

require provider and model, use temperature=1 and big tockens num for all gpt-5 models

pass openai api key with a window in gradio; pass openai key directly to the LLM agent constructor

allow a feedback question, remove 2nd prompt

add the temperature=1 and big context for gpt-5, as required by LiteLLM spec

remove unused openai api import

use dspy to abstract from model provider

vibe coded the cli and the llm

initial commit