Cracking the ChatGPT Puzzlebox

README.md

Cracking the ChatGPT Puzzlebox

I downloaded my chat history, now what? Maybe llm can help.

schema please

jq '.[0]' conversations.json | llm "extract schema"

I get a schema for each entry in the log. It's big!

what to index?

I want the 'title', 'conversation_id', and 'create_time'.

jq -c ".[] | {conversation_id, title, create_time}" conversations.json > index.jsonl

I put the 'conversation_id' first for llm embed-multi.

next steps

• use llm embed to create an embedding of the index.
• use llm similar to search the index
• write jq queries to extract an item by 'conversation_id'
• consider putting the whole log into sqlite. ??