Do you want to have several instances eg staging, production and dev and they can access different environment separately?
I think you can do this in several ways.
Like postman and other tools we provide https://docs.dify.ai/guides/workflow/variables you can define a variable in inputs and use if/else to distinguish the environments, or you can simply deploy multiple instances among the teams.

Not exactly. This is about custom tools, for instance ones that hit our own internal APIs. Consider the following:

• You wanna incorporate a custom tool that requests say user orders from your application.
• You add its OpenAPI spec, and save the custom tool.

Now you can use it, but the custom tool, during development, needs to hit localhost, but in production, it needs to hit our deployed application. How do you do this in a developer ergonomic way? I don't think Dify supports it.

Furthermore, fiddling with OpenAPI specification is itself not super ergonomic. I wonder if we can create custom tools via the API?

Do you have any suggestions? I think there might be a way that you could add 3 tools, and each of them would account for the environment. Eg, dev-tool-spec, stg-tool-spec, prod-tool-spec.

You can do that yeah, but then how do you keep them in sync? do you update all? and do so manually?

I think it would be ideal if Dify allows dynamic tool creation via an exposed API, similar to this for instance: https://humanloop.com/docs/v5/api-reference/tools/upsert.

I think you need a way that can get updated tools right?

https://humanloop.com/docs/v5/api-reference/tools/upsert

It can insert some specs I think, but it doesn't contain the implemtation.

We have a beta test plugin framework here. https://github.com/langgenius/dify-plugin-daemon However I can not talk much about it.

For example, it do have some kind of version controls