Reviewing the skill itself:
https://github.com/DataRecce/recce-claude-plugin/blob/main/plugins/recce-quickstart/commands/recce-setup.md
- If NOT installed: Tell the user they need Python 3.8+ installed and provide installation guidance for their OS.
Remember this is Python > 3.8, but also <= 3.12! see dbt-labs/dbt-core#12098
-
I don't love the git stash in there. What if they currently have interesting changes in the model that are not yet committed? It's really too easy to picture someone having a very mysterious time asking "I can see my changes, but recce can't!"
-
Would I have gotten the right result if I had not had a good profiles.yml? I think we might need an instruction that at least specifies the two outputs.
Reviewing the skill itself:
https://github.com/DataRecce/recce-claude-plugin/blob/main/plugins/recce-quickstart/commands/recce-setup.md
Remember this is Python > 3.8, but also <= 3.12! see dbt-labs/dbt-core#12098
I don't love the
git stashin there. What if they currently have interesting changes in the model that are not yet committed? It's really too easy to picture someone having a very mysterious time asking "I can see my changes, but recce can't!"Would I have gotten the right result if I had not had a good profiles.yml? I think we might need an instruction that at least specifies the two outputs.