LLMQuery

object

A query targeting the LLM usage analytics datasource.

* Additional properties are NOT allowed.
datasourcestringrequired

Allowed values:llm_usage

metricsarray[string]

List of aggregated metrics to collect across the requested time span.

Allowed values:total_tokensprompt_tokenscompletion_tokensai_request_countcostllm_cache_embeddings_latency_averagellm_cache_fetch_latency_averagellm_latency_averagellm_embeddings_tokensllm_embeddings_cost

Default:ai_request_count

dimensionsarray[string]

List of attributes or entity types to group by.

Allowed values:control_planecontrol_plane_groupgateway_serviceconsumerapplicationrouteai_providerai_response_modelai_request_modelllm_cache_statusllm_embeddings_providerllm_embeddings_modeltimerealmstatus_codestatus_code_groupedai_plugin

<= 2 items

filtersOne Of
array

A list of filters to apply to the query.

Filter by ai_pluginOne Of
Multiselect filtersobject
Show Child Parameters
granularitystring

Force time grouping into buckets of the specified duration. Only has an effect if “time” is in the “dimensions” list.

The granularity of the result may be coarser than requested. The finest allowed granularity depends on the query’s time range: data farther in the past may have coarser granularity. The exact result granularity will be reported in the response meta.granularity_ms field.

If granularity is not specified and “time” is in the dimensions list, a default will be chosen based on the time range requested.

Different relative times support different granularities:

  • 15m => tenSecondly, thirtySecondly, minutely
  • 1h => tenSecondly, thirtySecondly, minutely, fiveMinutely, tenMinutely
  • 6h => thirtySecondly, minutely, fiveMinutely, tenMinutely, thirtyMinutely, hourly
  • 12h => minutely, fiveMinutely, tenMinutely, thirtyMinutely, hourly
  • 24h => fiveMinutely, tenMinutely, thirtyMinutely, hourly
  • 7d => thirtyMinutely, hourly, twoHourly, twelveHourly, daily
  • 30d => hourly, twoHourly, twelveHourly, daily, weekly

For special time ranges:

  • current_week, previous_week => thirtyMinutely, hourly, twoHourly, twelveHourly, daily
  • current_month, previous_month => hourly, twoHourly, twelveHourly, daily, weekly

For absolute time ranges, daily will be used.

Allowed values:tenSecondlythirtySecondlyminutelyfiveMinutelytenMinutelythirtyMinutelyhourlytwoHourlytwelveHourlydailyweekly

time_rangeOne Of

The time range to query.

Default:{"type":"relative","time_range":"1h"}

Relative time rangeobject

A duration representing a relative-to-now span of time. Generally the start time is floored to the requested granularity. Eg 7d from now, with 1day granularity initiated at 2024-01-08T17:11:00+05:00 will query for the time range from 2024-01-01T00:00:00+05:00 to 2024-01-08T17:11:00+05:00. The exact start and end timestamps are returned in the result query in the meta.start and meta.end fields. If the granularity for the previous query was 1hour, it would query a time range from 2024-01-01T17:00:00+05:00 to 2024-01-08T17:11:00+05:00.

Show Child Parameters
Example

AgenticQuery

object

A query targeting the agentic usage analytics datasource.

* Additional properties are NOT allowed.
datasourcestringrequired

Allowed values:agentic_usage

metricsarray[string]

List of aggregated metrics to collect across the requested time span.

Allowed values:a2a_latency_averagea2a_response_size_sumerror_ratekong_latency_averagekong_latency_p50kong_latency_p95kong_latency_p99mcp_response_body_size_sumrequest_countrequest_per_minuterequest_size_averagerequest_size_p50request_size_p95request_size_p99request_size_sumresponse_latency_averageresponse_latency_p50response_latency_p95response_latency_p99response_size_averageresponse_size_p50response_size_p95response_size_p99response_size_sumupstream_latency_averageupstream_latency_p50upstream_latency_p95upstream_latency_p99

Default:request_count

dimensionsarray[string]

List of attributes or entity types to group by.

Allowed values:a2a_context_ida2a_errora2a_methoda2a_task_idapiapi_packageapi_productapi_product_versionapplicationconsumercontrol_planecontrol_plane_groupcountry_codedata_plane_nodedata_plane_node_versiongateway_servicemcp_errormcp_methodmcp_session_idmcp_tool_nameportalrealmresponse_sourceroutestatus_codestatus_code_groupedtimeupstream_status_codeupstream_status_code_grouped

<= 2 items

filtersOne Of
array

A list of filters to apply to the query.

Filter by a2a_context_idOne Of
Multiselect filtersobject
Show Child Parameters
granularitystring

Force time grouping into buckets of the specified duration. Only has an effect if “time” is in the “dimensions” list.

The granularity of the result may be coarser than requested. The finest allowed granularity depends on the query’s time range: data farther in the past may have coarser granularity. The exact result granularity will be reported in the response meta.granularity_ms field.

If granularity is not specified and “time” is in the dimensions list, a default will be chosen based on the time range requested.

Different relative times support different granularities:

  • 15m => tenSecondly, thirtySecondly, minutely
  • 1h => tenSecondly, thirtySecondly, minutely, fiveMinutely, tenMinutely
  • 6h => thirtySecondly, minutely, fiveMinutely, tenMinutely, thirtyMinutely, hourly
  • 12h => minutely, fiveMinutely, tenMinutely, thirtyMinutely, hourly
  • 24h => fiveMinutely, tenMinutely, thirtyMinutely, hourly
  • 7d => thirtyMinutely, hourly, twoHourly, twelveHourly, daily
  • 30d => hourly, twoHourly, twelveHourly, daily, weekly

For special time ranges:

  • current_week, previous_week => thirtyMinutely, hourly, twoHourly, twelveHourly, daily
  • current_month, previous_month => hourly, twoHourly, twelveHourly, daily, weekly

For absolute time ranges, daily will be used.

Allowed values:tenSecondlythirtySecondlyminutelyfiveMinutelytenMinutelythirtyMinutelyhourlytwoHourlytwelveHourlydailyweekly

time_rangeOne Of

The time range to query.

Default:{"type":"relative","time_range":"1h"}

Relative time rangeobject

A duration representing a relative-to-now span of time. Generally the start time is floored to the requested granularity. Eg 7d from now, with 1day granularity initiated at 2024-01-08T17:11:00+05:00 will query for the time range from 2024-01-01T00:00:00+05:00 to 2024-01-08T17:11:00+05:00. The exact start and end timestamps are returned in the result query in the meta.start and meta.end fields. If the granularity for the previous query was 1hour, it would query a time range from 2024-01-01T17:00:00+05:00 to 2024-01-08T17:11:00+05:00.

Show Child Parameters
Example

Tile

One Of
layoutobjectrequired

Information about how the tile is placed on the dashboard.

Examples:

  • a tile occupying the first half of the top row: { "position": { "col": 0, "row": 0 }, size: { "cols": 3, "rows": 1 } }
  • a tile occupying the second half of the top row: { "position": { "col": 3, "row": 0 }, size: { "cols": 3, "rows": 1 } }

Example:{"position":{"col":3,"row":0},"size":{"cols":3,"rows":1}}

* Additional properties are NOT allowed.
Show Child Parameters
typestringrequired

The type of tile. Chart tiles must have type ‘chart’.

Allowed values:chart

definitionobjectrequired

The tile’s definition, which consists of a query to fetch data and a chart to render the data.
Note that some charts expect certain types of queries to render properly. The documentation for the individual chart types has more information.

* Additional properties are NOT allowed.
Show Child Parameters
Example

ChartTile

object

A tile that queries data and renders a chart.

* Additional properties are NOT allowed.
layoutobjectrequired

Information about how the tile is placed on the dashboard.

Examples:

  • a tile occupying the first half of the top row: { "position": { "col": 0, "row": 0 }, size: { "cols": 3, "rows": 1 } }
  • a tile occupying the second half of the top row: { "position": { "col": 3, "row": 0 }, size: { "cols": 3, "rows": 1 } }

Example:{"position":{"col":3,"row":0},"size":{"cols":3,"rows":1}}

* Additional properties are NOT allowed.
Show Child Parameters
typestringrequired

The type of tile. Chart tiles must have type ‘chart’.

Allowed values:chart

definitionobjectrequired

The tile’s definition, which consists of a query to fetch data and a chart to render the data.
Note that some charts expect certain types of queries to render properly. The documentation for the individual chart types has more information.

* Additional properties are NOT allowed.
Show Child Parameters
Example

Dashboard

object

A JSON object describing a dashboard.

A dashboard is an array of tiles. All tiles are of type ‘chart’, which query data and render a chart displaying that data.

Dashboards have 6 columns and as many rows as necessary to display their tiles.

* Additional properties are NOT allowed.
tilesOne Of
arrayrequired

The array of tiles. We currently support up to 102 tiles, which is 17 rows of 6x1 tiles.

>= 0 characters<= 102 characters

ChartTileobject

A tile that queries data and renders a chart.

* Additional properties are NOT allowed.
Show Child Parameters
preset_filtersOne Of
array

An optional array of filters that are applied globally to all relevant tiles in the dashboard.

Whether or not a preset filter applies to a tile depends on the filter’s dimension. Some dimensions, like control_plane, are common to all datasources; other dimensions may only apply to one datasource.

Example:{"operator":"in","field":"control_plane","value":["00000000-0000-0000-0000-000000000000"]}{"operator":"in","field":"ai_request_model","value":["request_model1"]}

Filter by ai_pluginOne Of
Multiselect filtersobject
Show Child Parameters
Example