In Grafana Cloud, open the Adaptive Metrics plugin (Cost Management > Adaptive Metrics) and review the Recommendations tab, which lists high-cardinality metrics with suggested aggregation rules
For each recommended rule, examine which labels are proposed for removal (drop_labels) and verify no active dashboard or alert queries filter or group by those labels
Accept recommendations individually via the UI, or export all recommendations to a YAML/JSON rule file for review in code
Apply custom or accepted rules via the Adaptive Metrics HTTP API: fetch the current rules set, merge your changes, and PUT the updated rules array back to the API endpoint
Monitor the Active Series metric in Grafana Cloud to confirm cardinality reduction takes effect within 5–10 minutes of rule application
Manage rules at scale with the Adaptive Metrics Terraform provider (grafana_metrics_rules) to keep aggregation config version-controlled alongside your other IaC
Known gotchas
There is a 5–10 minute delay between uploading new aggregation rules and their effect appearing in active series counts; do not conclude a rule failed if the count does not drop immediately
Aggregated metrics lose the dropped labels permanently in the stored series; if you later need those labels for a new dashboard, you must re-ingest the original data — plan label drops carefully
The keep_labels parameter in the recommendations config prevents specific labels from ever being recommended for removal; use it for labels that are always required for alerting or SLO queries
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp