Hi Phoenix Support, I have a question in using phoenix for agent tracing. I鈥檓 tracing my agent application in Phoenix, and each node exchanges input and output as a JSON-formatted state. Phoenix is tracing that鈥攂ut does Phoenix have any functionality like pretty print or a typical JSON viewer, where I can expand and collapse each key? Or is that not supported?
Phoenix currently supports pretty-printing and compact display of JSON in the UI, including truncation and single-key collapsing, but it does not natively provide a full-featured expandable/collapsible JSON viewer (like tree view with expand/collapse per key) as of now. There are open feature requests and ongoing discussions to add such functionality, including syntax highlighting and collapsible subtrees, but this is not yet implemented in the main UI [github enhancement #8433].The current implementation uses components like CompactJSONCell and JSONText to render JSON in a compact or truncated form, and will show just the string if the JSON object has a single key, but does not support interactive expansion/collapse of nested keys [CompactJSONCell.tsx, JSONText.tsx]. Would you like more detail on the current rendering logic or the status of the enhancement request?Sources:
- github enhancement #8433
- CompactJSONCell.tsx
- JSONText.tsx
馃挕聽Hint: Mention RunLLM in the thread for followups.
Can you try updating the instrumentor to the latest version, i.e. pip install 'openinference-instrumentation-langchain>=0.1.49' The reason it鈥檚 not displaying nicely is that the objects were previously captured as strings.
Great. Thanks for testing! Yes, we unpack the object and serialized them into json
