Hi Guys, using Phoenix a lot and have a small feature request: As far as I can tell, there's no easy way to copy or export a trace (or parts of it) directly from the UI. This would be really useful when working on agentic workflows or tool-calling, where I often want to copy/paste a trace into an LLM for debugging or iteration. I know it鈥檚 possible to access traces via the DB / CLI / code, but having a UI option to export (e.g. Markdown or JSON) would be super helpful. Curious if I鈥檓 missing something or if this is on the roadmap.
You are correct: while Phoenix allows exporting traces and spans via the Python client and code (e.g., using get_spans_dataframe()), direct export of traces or parts of traces from the UI (such as to Markdown or JSON) is not currently a built-in feature. However, this enhancement has been requested by users and is acknowledged by the Phoenix team as a valuable addition, with related improvements under consideration or in progress for future releases. For now, exporting is primarily done programmatically or via dataset export features in the UI, but not as a one-click trace export in formats like Markdown or JSON from the UI itself. See discussions and documentation for more details: GitHub Issue #6332, UI Export Docs, Slack thread.Would you like more detail on current export workflows or updates on the feature roadmap?Sources:
- GitHub Issue #6332
- UI Export Docs
- Slack thread
馃挕聽Hint: Mention RunLLM in the thread for followups.
That sounds like a good idea to me, would you be open to creating a feature request here? https://github.com/Arize-ai/phoenix/issues The closest thing we have at the moment that I can think of from the UI is to:
select your span/trace via the checkbox on the table
click "add to dataset" in the toolbar that appears
click the "view" button in the notification that appears afterwards
download the entire dataset via the download button in the top right of the page, or, click the example row and copy the inputs or outputs
Are there particular fields beyond the input and output that you are interested? What kind of analysis would you like an LLM to do on them?
I will open the feature request, thanks. I am interested in all the spans, for example I would like to copy these. It's a tool call and the tool uses another llm call:
Made an issue here: https://github.com/Arize-ai/phoenix/issues/10746
