Are there good tips on debugging failed AI calls? ...

# koog-agentic-framework

[Nicolas Frenay]

Are there good tips on debugging failed AI calls?
In my use case I have some odd errors where Gemini is failing due to max tokens when running the structured output prompt, even though I shouldn't be hitting token limits.
To make it easier to see the prompts I instrumented the agent with OpenTelemetry + Langfuse locally, and built a koog version that doesn't hides the contents of the prompts.

[Konstantin]

Hi, @Nicolas Frenay
You may install Logging feature (example) or use OpenTelemetry.

[Nicolas Frenay]

thanks!