managing token limit

hit the token limit

system prompt needed improvement

Extract the string content from the LLM response object before passing it to `format_response_for_user`, and ensure it is a dictionary.

json is coming back

needs dict, not string

prompt is now a string, not a tuple, copilot nailed this one

json is not needed at all for this

vibe coding lead to a very lazy and wrong application of constrained decoding - simply cutting off the llm after a closing brace, bad bad bad - also I'm not sure why it was trying to deliver json at all, this plays to all of the weaknesses of gen ai, bad llm bad

better logging in key

it took out my imports

improved system prompt and container logs

.

initial commit

tag is not recognized

added tag

track 3 complete, demo video included

have to turn the mcp server on for it to work

included section in readme for the required demo videos even though they are not made yet

checkpoint 2 - we are in a very good position right now, audio transcription is working, gradio page looks good, local llms are working, returning the consultation, the json that will be sent to the mcp client, and a debugging panel

no matching dist found

I'm so confused

improved seperation of concerns best practice in the code, added print statements for better understanding of what code is doing

app.py wrapper for Gradio waa very bad idea, regoranizing project for clarity, utils folder will be very import of separation of concerns

t-shooting poor performance

Move plugin manifest to root for static file serving

wrong url

no logo

Remove Git LFS tracking and reset file state

add well known plugin manifest and logo for mcp integration

ignore .gradio folder

checkpoint 1

Revert "1. Structure responses with both diagnoses and follow-up questions"

1. Structure responses with both diagnoses and follow-up questions

see a progress counter when submit is it

submit button wasn't working

these integrations could potentially multiply winnings

actually... the gradio live link would allow me to show o4-mini-high the results, where I could not before

no need for gradio live

included badly needed links to where I'm developing

tags chosen highlight technical strengths and practical applications, I am attempting to target multiple prize categories, there's a limit to the number of tags

solved ffmpeg-related error

1. Removed global pipeline initialization to prevent warm-up crashes

FutureWarning: The input name `inputs` is deprecated. Please make sure to use `input_features` instead.

preventing segmentation fault

1. Added memory management

1. Added text input processing function

attention mask is not set

audio processing pipeliine

1. Removed unsafe `.get()` calls