What problem does it solve? When running a self-improvement loop for AI skills, you cannot tell whether telemetry is actually collecting evidence of skill usage. This Skill diagnoses the telemetry state so you know whether skills are being observed before trusting curator recommendations. ## Core Features & Use Cases - Telemetry State Diagnosis: Reports whether Skill telemetry is uninitialized, unavailable, no-op, or enabled, with a concrete remediation for each state. - Usage Event Inspection: When enabled, lists observed skills, recent events, and outcome counts from partitioned rollups. - Pre-Curator Check: Run before the report-only Skill curator to confirm evidence is being collected. - Use Case: Before asking the curator which skills to archive, run this diagnostic to verify graph mode with skill telemetry is active and events exist. ## Quick Start Ask the agent to check the Skill telemetry status and tell you whether skill usage is being observed.