Replies: 2 comments 2 replies
-
|
Auto-instrumentation with openinference-google-adk currently does not fully trace Google ADK artifacts created from bytes (like images or PDFs). This is a known limitation: the instrumentation only extracts text fields from To trace artifact operations (such as Here’s a quick example for manual tracing: from phoenix.trace import tracer
with tracer.start_as_current_span("save_artifact") as span:
span.set_attribute("artifact.filename", filename)
span.set_attribute("artifact.mime_type", mime_type)
span.set_attribute("artifact.data_size", len(data))
# ... perform save_artifact logic ...
span.set_status(Status(StatusCode.OK))If you want to work around the current instrumentation limitation for image/PDF parts, you can monkey-patch the For best results, run your agent in a fresh environment, initialize the instrumentor only once, and use a Let me know if you need a more detailed code template for manual instrumentation! To reply, just mention @dosu. How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other |
Beta Was this translation helpful? Give feedback.
-
|
Hey @ruqianq, I'm not deeply familiar with Google ADK. Can you take a look the linked issue to see if it describes the issue you are facing? |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Hi community,
I use auto-instrumentation from openinference and from phoenix UI I dont see the artifact is being traced on? do I need to manually trace on artifact, what attributes would you recommend me to use? Thanks!
for reference: https://google.github.io/adk-docs/artifacts/
Beta Was this translation helpful? Give feedback.
All reactions