Problem
Users can see agent usage and limits, but they lack visibility into when skills are invoked. A customer said skills are valuable but it is difficult to tell when they fired, which makes their behavior hard to trust and debug.
Desired outcome
Add skill-level observability to Usage & Limits, including:
- which skill was invoked
- when it ran
- which pane, session, and agent invoked it
- whether it completed or failed
- a way to inspect the invocation details without exposing secrets
If reliable skill invocation requires explicit instrumentation rather than inference from prompts or output, capture that distinction in the event model.
Source: customer feedback call, September 4, 2026.
Problem
Users can see agent usage and limits, but they lack visibility into when skills are invoked. A customer said skills are valuable but it is difficult to tell when they fired, which makes their behavior hard to trust and debug.
Desired outcome
Add skill-level observability to Usage & Limits, including:
If reliable skill invocation requires explicit instrumentation rather than inference from prompts or output, capture that distinction in the event model.
Source: customer feedback call, September 4, 2026.