# Nushell core team meeting 2025-06-25

## Attendees
- Stefan
- Rose
- Piet
- Michael
- Bahex
- Douglas
- Jan9103
- Andy
- Wind
- Jack
- Darren, 45 minutes late

## Agenda
- [x] How do we want to deal with AI in our PRs?
- [ ] Lets talk about the Sets subject
- [x] Macros; When do we allow? When do we disallow? These seem overkill to me (darren) [PR 16001](https://github.com/nushell/nushell/pull/16001/files) for a `random dice --sides 0` fix but maybe I'm wrong. Help me understand.
- [x]

## Discussed Topics

### AI use in PRs/issues

- Piet: use AI and manual review fine, descriptions/responses should reflect human behind it
- let's hash out our expectations in a hackmd document

### experimental options ~ feature gates
- https://github.com/nushell/nushell/pull/16028
- Bikeshedding
- experimental options
- feature flags (collides with the compile time)
- experimental features (very verbose)
- `Stability` enum
- bikeshed the names
- don't call things stable

### macro use or more specifically [PR 16001](https://github.com/nushell/nushell/pull/16001/files)
- from the internal chat: macros generally it depends
- specific to the PR don't overgeneralize if there are no uses
- Darren: try to split API development from local feature development for better unterstanding.

### set PR
- we didn't respond
- general consensus we don't like new syntax for that
- for the type, we also don't have a majority so we would close and prefere experimentation in plugin or standard library for now.