Some members of our extended team had a real struggle getting started with Moonshot via Python recently, because the basic notebook tutorial and the Moonshot docs collectively:
Don't really make it clear enough what moonshot-data is, how it relates to Moonshot, and what installing it does.
Don't signpost enough about how connectors & endpoints work, for somebody to adapt & follow the tutorial if they don't have easy access to an OpenAI key and want to use a different provider.
It feels like the docs website is missing a high-level conceptual guide explaining the abstractions Moonshot defines e.g. Connector, Endpoint, Recipe, Cookbook, etc?
To be honest, I find the whole moonshot-data installation process clunky... If it's a simple git clone, what value do we get by hiding that behind a python -m moonshot -i moonshot-data CLI tool? Seems like it would be better to openly tell users what this step is doing and how moonshot references the assets?
I can propose some edits for 1 in a PR, but 2 & 3 feel like bigger changes...
Some members of our extended team had a real struggle getting started with Moonshot via Python recently, because the basic notebook tutorial and the Moonshot docs collectively:
moonshot-datais, how it relates to Moonshot, and what installing it does.I suggest:
python -m moonshot -i moonshot-dataCLI tool? Seems like it would be better to openly tell users what this step is doing and how moonshot references the assets?I can propose some edits for 1 in a PR, but 2 & 3 feel like bigger changes...