Amine S.

Hey everyone 👋 I'm working on adding custom tracing for Agno workflows using OpenInference. It’s been smooth so far—instrumenting agents and tools is straightforward by subclassing BaseInstrumentor and extending the wrappers provided in the Agno module (like _FunctionCallWrapper). However, I’ve hit a challenge when trying to trace Agno’s KnowledgeTools methods (search, think, analyze). Since there’s no existing wrapper for KnowledgeTools in OpenInference, it’s tricky to emit structured spans, especially when I want to capture and inspect the retrieved document chunks. I’ve opened an issue here with more context: https://github.com/Arize-ai/openinference/issues/1770 Does anyone have suggestions—or even a hacky workaround—for tracing KnowledgeTools properly until a native wrapper is available? Thanks in advance!

Hi everyone! 👋 Hope you're all doing well. We've recently rolled out Phoenix in production and it's been an awesome experience so far! I'm now working on a project with Agno and using their team config. One challenge I’ve run into is that the trace logs can get quite verbose—especially due to their formatting and reasoning steps, which makes multistep traces a bit tricky to navigate. Is there a recommended way in Phoenix to control or filter the trace data more selectively? I’m particularly interested in logging or visualizing only the most relevant traces to our workflow. Would really appreciate any tips or best practices for keeping trace logs clean and easy to work with. Thanks!