The project's metrics store

The metrics store keeps the service's metrics as time series and evaluates its alert and SLO rules. Which one it is depends on the project's platform, so it is found in its configuration rather than assumed.

Finding it

• The monitoring config: scrape targets (prometheus.yml), recording and alert rule files, dashboards kept in the repository.
• SLO definitions, if any: the specs that generate the SLO's recording and alert rules.
• How the agent reaches it: the HTTP API (/api/v1/query and /api/v1/query_range in Prometheus and the stores compatible with it), promtool, or an MCP server.

Using it

• Query over the HTTP API with PromQL: curl -G <server>/api/v1/query --data-urlencode 'query=<expr>'.
• Compare versions by a label, over the same window and the same traffic.
• Check that a query returns series before trusting its value: an empty result is no data, not zero.
• Check rules before they ship: promtool check rules <file> and promtool test rules <file>.

Examples: Prometheus (PromQL), VictoriaMetrics; Sloth to generate SLO rules; Alertmanager.