Thinking in the open about strategy for documentation injection: Could use 'fuzzy-finder'-esque wrapper to determine concepts or words that are relevant to a user's query. Might have to start with dict of common associations. Ex:

['DSP', 'Data Security Platform' ...] --> ['$def_of_dsp', $opt_link_to_readme_section]
['PEP', 'Policy Enforcement Points' ...] --> ['$def_of_peps', $opt_link_to_readme_section]
...

This, along with basic sanitization wrapping in sanitizer.go would be the basis for entire prompt. Open item would be to adjust parameters based on the token_lens/context_size of the model when considering tradeoffs for speed... default is 2K, 8K IIUC.

@andrewrust-virtru when considering the reducing context have you looked at prompt-engineering with llama2? https://medium.com/@eboraks/llama-2-prompt-engineering-extracting-information-from-articles-examples-45158ff9bd23

For instance if we append <SYS>...</SYS> to the user's prompt we can ensure the response will be limited by what we want.

<s>[INST] <<SYS>>
You are a helpful, respectful and honest assistant. Always answer as helpfully as possible, while being safe.  
Your answers should not include any harmful, unethical, racist, sexist, toxic, dangerous, or illegal content. 
Please ensure that your responses are socially unbiased and positive in nature.
If a question does not make any sense, or is not factually coherent, explain why instead of answering something not correct. 
If you don't know the answer to a question, please don't share false information.
<</SYS>>

Tell me how to choose a favorite color

UPDATE(08/06/24): https://github.com/virtru-corp/data-security-platform/issues/455#issuecomment-2272393715

In preparation for intern presentation Thursday, ending 'new feature' commits. Just necessary bug fixes for a working demo. Updating documentation for OpenTDF issue #455 on base issue for otdfctl: issue #12.

Related to archived Atlassian goal and abandoned otdfctl branch #40