Hi @nikitathakur ,
This is an interesting use case and I totally get why you’d want to have all that monitoring and governance for Fabric data agents. From what I know, out-of-the-box Fabric doesn’t provide a full suite for deep agent monitoring like trust scores, detailed user feedback, or agent-specific performance logs directly in the UI. Some high-level metrics and basic logs are available, but they’re mostly focused on overall system activity, not granular agent behavior.
A couple of things you could try though:
- If your agents are running queries via Fabric’s native tools (like Data Activator or pipelines), you can pull operation logs and maybe combine them with custom logging in your agent logic. This way, you might get things like agent name, timestamps, and error events, but you’d probably need to stitch together dashboards manually (Power BI is good for this).
- For user feedback or capturing errors, there’s no built-in feedback/annotation loop yet, so you’d have to build a small feedback form or logging mechanism on top of your solution.
- Trust scores and model quality would likely require custom metrics. You might consider tagging logs with quality indicators or using external monitoring frameworks to track these over time.
If you want detailed governance, it’s mostly a matter of combining what’s available in Fabric logs with your own custom tracking and visualization. There are some docs on Fabric monitoring but nothing super deep on agent-level telemetry as of now.
If my response resolved your query, kindly mark it as the Accepted Solution to assist others. Additionally, I would be grateful for a 'Kudos' if you found my response helpful.
