Now assist custom skill keeps getting the output sliced

[PedroMigueG8564]

Hello dear community,

I'm developing a custom now assist skill where it gets fed by inputs in the format of JSON Array.

While getting the response, the result output gets sliced down in random way, so for the same input, it will get sliced in different parts. Seems like the response isn't being properly build into the output.

Does anyone get this issue too? Also, because of this, the response keeps getting different from test to test with the same input.

Is there anything I can do to prevent the output from being sliced?

Thank you for your time!

[Vamsi_Krishna07]

Hi @PedroMigueG8564 ,

You can check what is happening in the sys_generative_ai_log table. It logs all Now Assist–related transactions, including the prompt and the response received. In your case, the response may be exceeding the token limit, which is likely why it is getting truncated.