Even when using the tool builder in the playground which sets things to strict: true always My responses are always logged in the openai dashboard chat completions logger as strict: false.
I’m looking into this, because I’'m not confident that my tool schemas are correct and there’s no linter that I am aware of. I was hoping I could use the dashboard logs as a sanity check because I’m seeing some values being generated that are not present in the enum fields that I have selected.
Has anyone else noticed this? How do I verify that my tool is actually in strict mode?