List monitors
List monitors the caller can read, with filtering and cursor-based pagination. Results are ordered by creation time, newest first. Deleted and draft monitors are excluded.
The shape of each returned monitor varies by type (DATA_QUALITY,
PERFORMANCE, DRIFT, CUSTOM_METRIC, TRACING).
All filters are optional and compose with AND semantics (a monitor
must match every provided filter). When a filter is omitted, no
filtering is applied for that field. Filters that cannot be satisfied
together — a metric filter from one family combined with a type from
another, or two metric filters from different families — are valid
input and return a 200 with an empty monitors array, as does any
other combination that simply matches nothing.
The four name filters do not all match the same way. name and
space_name are case-insensitive substring searches, so name=prod
matches “production”. project_name and dimension_name are exact,
case-sensitive matches, so they need the full name as stored — for
dimension_name, the value copied verbatim from a returned monitor’s
dimension.name.
An identifier that is not a well-formed ID of the expected kind
returns a 400. A well-formed space_id, project_id, or
custom_metric_id that either does not exist or is not readable by
the caller returns the same 404 in both cases, so the response never
reveals whether the referenced resource exists.
A caller whose credentials grant monitor read access in no space at
all receives a 403.
Authorizations
Most Arize AI endpoints require authentication. For those endpoints that require authentication, include your API key in the request header using the format
Query Parameters
Filter search results to a particular space ID A universally unique identifier (base64-encoded opaque string).
"RW50aXR5OjEyMzQ1"
Case-insensitive substring filter on the space name. Narrows results to resources in spaces whose name contains the given string. If omitted, no space name filtering is applied and all resources are returned.
255Case-insensitive substring filter on the resource name. Returns only
resources whose name contains the given string. For example,
name=prod matches "production", "my-prod-dataset", etc. If omitted,
no name filtering is applied and all resources are returned.
255Filter results to resources associated with a specific project (base64 identifier). If omitted, results are not filtered by project.
A universally unique identifier (base64-encoded opaque string).
"RW50aXR5OjEyMzQ1"
Exact-match filter on the name of the project the monitor's primary
metric is computed over. Unlike name and space_name, this is an
exact (case-sensitive) match, not a substring search. If omitted, no
project name filtering is applied.
255Filter by monitor type. Types are exact: DATA_QUALITY does not
include TRACING monitors, and PERFORMANCE does not include
CUSTOM_METRIC monitors. If omitted, monitors of all types are
returned.
The kind of monitor. Determines which type-specific fields apply.
DATA_QUALITY- Monitors a data quality metric (e.g. percent empty, CARDINALITY).PERFORMANCE- Monitors a model performance metric (e.g. ACCURACY, RMSE).DRIFT- Monitors distributional drift of a feature/output.CUSTOM_METRIC- Monitors a user-defined custom metric.TRACING- Monitors a span/trace-derived metric (e.g. span attributes, evals).
DATA_QUALITY, PERFORMANCE, DRIFT, CUSTOM_METRIC, TRACING Filter by the monitor's current evaluation state (TRIGGERED,
CLEARED, or NO_DATA). If omitted, monitors in every state are
returned.
The monitor's current state from its most recent evaluation.
TRIGGERED- The metric breached the threshold.CLEARED- The metric is within healthy bounds.NO_DATA- No data was available to evaluate.
TRIGGERED, CLEARED, NO_DATA Filter by whether notifications fire on a triggered transition.
true returns only monitors with notifications enabled; false
returns only monitors with notifications disabled. If omitted,
monitors are returned regardless of notification state.
Filter to monitors whose metric is computed over a dimension of this
category. Values copied from a returned monitor's dimension.category
work as filters. If omitted, no dimension category filtering is
applied.
The category of dimension the metric is evaluated over.
FEATURE_LABEL, TAG, MODEL_VERSION, BATCH_ID, SPAN_ATTRIBUTE, LLM_EVAL, USER_ANNOTATION, ACTUALS, ACTUALS_SCORE, ACTUALS_CLASS, PREDICTIONS, PREDICTIONS_SCORE, PREDICTIONS_CLASS Exact-match filter on the name of the dimension the monitor's metric
is computed over. Values copied from a returned monitor's
dimension.name work as filters. If omitted, no dimension name
filtering is applied.
255Filter to monitors computing this data quality metric. Matches both
DATA_QUALITY and TRACING monitors; combine with type to narrow
to one of them. If omitted, no data quality metric filtering is
applied.
The data quality metric computed over the selected dimension.
COUNT, PERCENT_EMPTY, CARDINALITY, NEW_VALUES, MISSING_VALUES, AVG, SUM, STANDARD_DEVIATION, P50, P95, P99, P99_9, AVERAGE_STRING_LIST_LENGTH Filter to PERFORMANCE monitors computing this performance metric.
Does not match CUSTOM_METRIC monitors. If omitted, no performance
metric filtering is applied.
The model performance metric.
ACCURACY, CALIBRATION, F_1, FALSE_NEGATIVE_DENSITY, FALSE_NEGATIVE_RATE, FALSE_POSITIVE_RATE, PRECISION, RECALL, SENSITIVITY, SPECIFICITY, MICRO_AVERAGED_PRECISION, MACRO_AVERAGED_PRECISION, MULTI_CLASS_PRECISION, MICRO_AVERAGED_RECALL, MACRO_AVERAGED_RECALL, MULTI_CLASS_RECALL, ACTUALS_AVERAGE, MAE, MAPE, MEAN_ERROR, MSE, PREDICTION_AVERAGE, RMSE, R_SQUARED, SMAPE, WAPE, MASE, AUC, LOG_LOSS, NDCG, PR_AUC, GROUP_AUC, MRR, RANKING_MAP, RANKING_RECALL, RANKING_PRECISION, DISPARATE_IMPACT, FALSE_POSITIVE_RATE_PARITY, RECALL_PARITY Filter to DRIFT monitors computing this drift metric. If omitted,
no drift metric filtering is applied.
The statistical drift metric. PSI/KL/JS/KS apply to structured data;
EUCLIDEAN_DISTANCE/COSINE_SIMILARITY apply to unstructured (embedding) data.
PSI, KL, JS, KS, EUCLIDEAN_DISTANCE, COSINE_SIMILARITY Filter to CUSTOM_METRIC monitors evaluating this custom metric
(base64 identifier). If omitted, no custom metric filtering is
applied.
A universally unique identifier (base64-encoded opaque string).
"RW50aXR5OjEyMzQ1"
Maximum items to return. Defaults to 50 if omitted; maximum is 100.
1 <= x <= 100Opaque pagination cursor returned from a previous response
(pagination.next_cursor). Treat it as an unreadable token; do not
attempt to parse or construct it.
Response
Returns a list of monitor objects