What it does
Protected metrics and retention rules remain in the collector, making semantic judgment one input to telemetry policy rather than the policy itself.
Integrations
An experimental OpenTelemetry processor that judges metric metadata before local policy annotates, reduces or drops instruments.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
Protected metrics and retention rules remain in the collector, making semantic judgment one input to telemetry policy rather than the policy itself.
For a server with too many measurements to store, first try the mode that rates them without removing any. Ask a developer to compare the AI ratings with data needed to find known problems. Names and descriptions are sent to an outside AI service. Check what you may share before starting, and do not let an untested rating decide what gets deleted.