Translate

Start from a customer APM / infra / business metric, get the log patterns whose rate curves move with it. Correlate pointed in one direction (metric → patterns), so the agent doesn't have to specify the anchor type.

Example

"log patterns behind apm.payments.latency?"

Payment_Gateway_Timeout (r=0.91) · DB_Connection_Refused (r=0.83) · Retry_Exhausted (r=0.71). All three share the service label with the metric, so the matches are structurally grounded — not just timing coincidence.

More to ask

• "patterns co-moving with container memory in pod-X"
• "what logs explain this APM anomaly?"
• "patterns behind the p99 latency spike on payments-svc"

Prerequisites

Same as Correlate.

Tool schema (advanced)

Field	Type	Required	Default	Description
customer_metric	string	yes	—	Customer PromQL metric expression.
window	string	no	1h	Time window for correlation. Alias: timeRange.
timeRange	string	no	—	Alias for window for consistency with the other tools.
depth	string	no	normal	shallow / normal / deep.
step	string	no	60s	Bucket step.
minimum_confidence	number	no	0.3	Minimum combined confidence. 0–1.
environment	string	no	—	Environment nickname.