Agent Metric Monitor

string · required

The agent whose spans the monitor reads. Agent monitors name an agent instead of a table; the warehouse source is derived from it. The value takes one of two forms, told apart by its shape:

• Warehouse platform agents (e.g. Snowflake Cortex, Databricks) — a <database>:<schema>.<name> reference, where <name> is the agent's display name when it has one (otherwise its underlying identifier). No trace_table.
• OpenTelemetry agents — the agent's bare service_name. By default the OpenTelemetry trace store is resolved automatically; set trace_table only when the spans live in a specific warehouse table.

A value with a : and a dotted tail (db:schema.name) is read as a platform reference; a bare value is an OpenTelemetry service_name.

agent: "ANALYTICS:AGENTS.support_agent"

string · optional

<database>:<schema>.<table> naming the warehouse trace table that holds an OpenTelemetry agent's spans. Required when the spans live in your warehouse, or when the warehouse holds more than one OpenTelemetry trace table. Omit it for platform agents and for agents resolved from the default OpenTelemetry trace store.

trace_table: "ingest:opentelemetry.traces"

array of objects · required

Same shape as the Metric Monitor — a metric, optional fields, an operator (AUTO, GT, LT, range operators, …), and thresholds. See Available Metrics for metric names.

alert_conditions:
  - metric: NULL_RATE
    fields: [prompts]
    operator: AUTO

array of objects · optional

Refines the monitor to specific spans. The agent is already set by agent, so it need not be repeated. Each entry sets one or more of workflow, task, span_name, each an object with a value.

agent_span_filters:
  - workflow:
      value: checkout

enum · optional

Accepted values: HOUR · DAY · WEEK · MONTH (uppercase)

Buckets metrics by the chosen interval.

aggregate_by: HOUR

boolean · optional · mutually exclusive

Raise the scoring grain from individual spans to whole traces (is_agent_trace_aggregation) or whole conversations (is_agent_conversation_aggregation). Setting both fails validation. Include the flag in the template to re-apply at the same grain — omitting it reverts to span grain.

is_agent_trace_aggregation: true

enum · optional

Accepted values: low · medium · high

Tunes how lax the ML-generated thresholds are for AUTO alert conditions.

sensitivity: medium

array of strings · optional

Segment metrics by up to 5 fields (segment_fields) or one SQL expression (segment_sql). Use one or the other, or neither.

segment_fields:
  - model_name

object · optional

Same shape as the Metric Monitor — type (fixed, dynamic, manual), interval_minutes, start_time, timezone.

schedule:
  type: fixed
  interval_minutes: 60
  start_time: "2025-01-01T00:00:00+00:00"

The shared monitor envelope works the same as other monitor types:

• name string · required — unique identifier in the namespace; renaming creates a new monitor.
• description string · required — max 512 characters.
• warehouse string · optional — UUID or name; overrides the montecarlo.yml default.
• connection_name string · optional — query engine to use within the warehouse.
• notes string · optional — shown in the UI, not in notifications.
• audiences / failure_audiences array of strings · optional — notification channels.
• priority enum · optional — P1–P5.
• tags array of objects · optional — name (required) + value (optional).
• data_quality_dimension enum · optional — ACCURACY · COMPLETENESS · CONSISTENCY · TIMELINESS · UNIQUENESS · VALIDITY.
• domains array of strings · optional (required on accounts created after January 2025) — domain for the monitor.