Would you be open to me adding an eval on the inputs to custom queries? It would make queries even more flexible.
Here are two examples of custom inputs I’d like to have. Right now, having access to those inputs requires adding new cases to resolve-input, i.e. a full code change. It’d be really cool if it could all just live in the users’ markdown files and be more dynamic!
There’s already an eval on
result-transform, so I don’t think it’d affect the security model.