There are some evals in the repo, already, that are using waza.
I'd like to examine those a bit more - it's an example of the kind of test where I just need to know that it would have called into or used 'azd', without having to run the actual action. These aren't testing azd's AI features, but more just that normal LLMs are able to pick azd when they have to accomplish a task.
Given there's a lot of them in there (https://github.com/Azure/azure-dev/tree/main/cli/azd/test/eval), and some of them explicitly mention 'azd' in their prompt, we might need to do some changes.
There are some evals in the repo, already, that are using waza.
I'd like to examine those a bit more - it's an example of the kind of test where I just need to know that it would have called into or used 'azd', without having to run the actual action. These aren't testing azd's AI features, but more just that normal LLMs are able to pick azd when they have to accomplish a task.
Given there's a lot of them in there (https://github.com/Azure/azure-dev/tree/main/cli/azd/test/eval), and some of them explicitly mention 'azd' in their prompt, we might need to do some changes.