MCP Tools Reference: monitoring.googleapis.com

Tool: list_metric_descriptors

Use this as the primary tool to discover the types of metrics available in a Google Cloud project. This is a good first step to understanding what data is available for monitoring and building dashboards or alerts.

The following sample demonstrate how to use curl to invoke the list_metric_descriptors MCP tool.

Curl Request 
                  
curl --location 'https://monitoring.googleapis.com/mcp' \
--header 'content-type: application/json' \
--header 'accept: application/json, text/event-stream' \
--data '{
  "method": "tools/call",
  "params": {
    "name": "list_metric_descriptors",
    "arguments": {
      // provide these details according to the tool's MCP specification
    }
  },
  "jsonrpc": "2.0",
  "id": 1
}'

Input Schema

The ListMetricDescriptors request.

ListMetricDescriptorsRequest

JSON representation 
{
  "name": string,
  "filter": string,
  "pageSize": integer,
  "pageToken": string,
  "activeOnly": boolean
}
Fields
name

string

Required. The project on which to execute the request. The format is:

projects/[PROJECT_ID_OR_NUMBER]
filter

string

Optional. If this field is empty, all custom and system-defined metric descriptors are returned. Otherwise, the filter specifies which metric descriptors are to be returned. For example, the following filter matches all custom metrics:

metric.type = starts_with("custom.googleapis.com/")
pageSize

integer

Optional. A positive number that is the maximum number of results to return. The default and maximum value is 10,000. If a page_size <= 0 or > 10,000 is submitted, will instead return a maximum of 10,000 results.
pageToken

string

Optional. If this field is not empty then it must contain the nextPageToken value returned by a previous call to this method. Using this field causes the method to return additional results from the previous method call.
activeOnly

boolean

Optional. If true, only metrics and monitored resource types that have recent data (within roughly 25 hours) will be included in the response. - If a metric descriptor enumerates monitored resource types, only the monitored resource types for which the metric type has recent data will be included in the returned metric descriptor, and if none of them have recent data, the metric descriptor will not be returned. - If a metric descriptor does not enumerate the compatible monitored resource types, it will be returned only if the metric type has recent data for some monitored resource type. The returned descriptor will not enumerate any monitored resource types.

Output Schema

The ListMetricDescriptors response.

ListMetricDescriptorsResponse

JSON representation 
{
  "metricDescriptors": [
    {
      object (MetricDescriptor)
    }
  ],
  "nextPageToken": string
}
Fields
metricDescriptors[]

object (MetricDescriptor)

The metric descriptors that are available to the project and that match the value of filter, if present.
nextPageToken

string

If there are more results than have been returned, then this field is set to a non-empty value. To see the additional results, use that value as page_token in the next call to this method.

MetricDescriptor

JSON representation 
{
  "name": string,
  "type": string,
  "labels": [
    {
      object (LabelDescriptor)
    }
  ],
  "metricKind": enum (MetricKind),
  "valueType": enum (ValueType),
  "unit": string,
  "description": string,
  "displayName": string,
  "metadata": {
    object (MetricDescriptorMetadata)
  },
  "launchStage": enum (LaunchStage),
  "monitoredResourceTypes": [
    string
  ]
}
Fields
name

string

The resource name of the metric descriptor.
type

string

The metric type, including its DNS name prefix. The type is not URL-encoded. All user-defined metric types have the DNS name custom.googleapis.com or external.googleapis.com. Metric types should use a natural hierarchical grouping. For example:

"custom.googleapis.com/invoice/paid/amount"
"external.googleapis.com/prometheus/up"
"appengine.googleapis.com/http/server/response_latencies"
labels[]

object (LabelDescriptor)

The set of labels that can be used to describe a specific instance of this metric type. For example, the appengine.googleapis.com/http/server/response_latencies metric type has a label for the HTTP response code, response_code, so you can look at latencies for successful responses or just for responses that failed.
metricKind

enum (MetricKind)

Whether the metric records instantaneous values, changes to a value, etc. Some combinations of metric_kind and value_type might not be supported.
valueType

enum (ValueType)

Whether the measurement is an integer, a floating-point number, etc. Some combinations of metric_kind and value_type might not be supported.
unit

string

The units in which the metric value is reported. It is only applicable if the value_type is INT64, DOUBLE, or DISTRIBUTION. The unit defines the representation of the stored metric values.

Different systems might scale the values to be more easily displayed (so a value of 0.02kBy might be displayed as 20By, and a value of 3523kBy might be displayed as 3.5MBy). However, if the unit is kBy, then the value of the metric is always in thousands of bytes, no matter how it might be displayed.

If you want a custom metric to record the exact number of CPU-seconds used by a job, you can create an INT64 CUMULATIVE metric whose unit is s{CPU} (or equivalently 1s{CPU} or just s). If the job uses 12,005 CPU-seconds, then the value is written as 12005.

Alternatively, if you want a custom metric to record data in a more granular way, you can create a DOUBLE CUMULATIVE metric whose unit is ks{CPU}, and then write the value 12.005 (which is 12005/1000), or use Kis{CPU} and write 11.723 (which is 12005/1024).

The supported units are a subset of The Unified Code for Units of Measure standard:

Basic units (UNIT)

  • bit bit
  • By byte
  • s second
  • min minute
  • h hour
  • d day
  • 1 dimensionless

Prefixes (PREFIX)

  • k kilo (10^3)
  • M mega (10^6)
  • G giga (10^9)
  • T tera (10^12)
  • P peta (10^15)
  • E exa (10^18)
  • Z zetta (10^21)
  • Y yotta (10^24)

  • m milli (10^-3)

  • u micro (10^-6)
  • n nano (10^-9)
  • p pico (10^-12)
  • f femto (10^-15)
  • a atto (10^-18)
  • z zepto (10^-21)
  • y yocto (10^-24)

  • Ki kibi (2^10)

  • Mi mebi (2^20)
  • Gi gibi (2^30)
  • Ti tebi (2^40)
  • Pi pebi (2^50)

Grammar

The grammar also includes these connectors:

  • / division or ratio (as an infix operator). For examples, kBy/{email} or MiBy/10ms (although you should almost never have /s in a metric unit; rates should always be computed at query time from the underlying cumulative or delta value).
  • . multiplication or composition (as an infix operator). For examples, GBy.d or k{watt}.h.

The grammar for a unit is as follows:

Expression = Component { "." Component } { "/" Component } ;

Component = ( [ PREFIX ] UNIT | "%" ) [ Annotation ]
          | Annotation
          | "1"
          ;

Annotation = "{" NAME "}" ;

Notes:

  • Annotation is just a comment if it follows a UNIT. If the annotation is used alone, then the unit is equivalent to 1. For examples, {request}/s == 1/s, By{transmitted}/s == By/s.
  • NAME is a sequence of non-blank printable ASCII characters not containing { or }.
  • 1 represents a unitary dimensionless unit of 1, such as in 1/s. It is typically used when none of the basic units are appropriate. For example, "new users per day" can be represented as 1/d or {new-users}/d (and a metric value 5 would mean "5 new users). Alternatively, "thousands of page views per day" would be represented as 1000/d or k1/d or k{page_views}/d (and a metric value of 5.3 would mean "5300 page views per day").
  • % represents dimensionless value of 1/100, and annotates values giving a percentage (so the metric values are typically in the range of 0..100, and a metric value 3 means "3 percent").
  • 10^2.% indicates a metric contains a ratio, typically in the range 0..1, that will be multiplied by 100 and displayed as a percentage (so a metric value 0.03 means "3 percent").
description

string

A detailed description of the metric, which can be used in documentation.
displayName

string

A concise name for the metric, which can be displayed in user interfaces. Use sentence case without an ending period, for example "Request count". This field is optional but it is recommended to be set for any metrics associated with user-visible concepts, such as Quota.
metadata

object (MetricDescriptorMetadata)

Optional. Metadata which can be used to guide usage of the metric.
launchStage

enum (LaunchStage)

Optional. The launch stage of the metric definition.
monitoredResourceTypes[]

string

Read-only. If present, then a time series, which is identified partially by a metric type and a MonitoredResourceDescriptor, that is associated with this metric type can only be associated with one of the monitored resource types listed here.

LabelDescriptor

JSON representation 
{
  "key": string,
  "valueType": enum (ValueType),
  "description": string
}
Fields
key

string

The key for this label. The key must meet the following criteria:

  • Does not exceed 100 characters.
  • Matches the following regular expression: [a-zA-Z][a-zA-Z0-9_]*
    • The first character must be an upper- or lower-case letter.
    • The remaining characters must be letters, digits, or underscores.
valueType

enum (ValueType)

The type of data that can be assigned to the label.
description

string

A human-readable description for the label.

MetricDescriptorMetadata

JSON representation 
{
  "launchStage": enum (LaunchStage),
  "samplePeriod": string,
  "ingestDelay": string,
  "timeSeriesResourceHierarchyLevel": [
    enum (TimeSeriesResourceHierarchyLevel)
  ]
}
Fields
launchStage
(deprecated)

enum (LaunchStage)

Deprecated. Must use the MetricDescriptor.launch_stage instead.
samplePeriod

string (Duration format)

The sampling period of metric data points. For metrics which are written periodically, consecutive data points are stored at this time interval, excluding data loss due to errors. Metrics with a higher granularity have a smaller sampling period.

A duration in seconds with up to nine fractional digits, ending with 's'. Example: "3.5s".
ingestDelay

string (Duration format)

The delay of data points caused by ingestion. Data points older than this age are guaranteed to be ingested and available to be read, excluding data loss due to errors.

A duration in seconds with up to nine fractional digits, ending with 's'. Example: "3.5s".
timeSeriesResourceHierarchyLevel[]

enum (TimeSeriesResourceHierarchyLevel)

The scope of the timeseries data of the metric.

Duration

JSON representation 
{
  "seconds": string,
  "nanos": integer
}
Fields
seconds

string (int64 format)

Signed seconds of the span of time. Must be from -315,576,000,000 to +315,576,000,000 inclusive. Note: these bounds are computed from: 60 sec/min * 60 min/hr * 24 hr/day * 365.25 days/year * 10000 years
nanos

integer

Signed fractions of a second at nanosecond resolution of the span of time. Durations less than one second are represented with a 0 seconds field and a positive or negative nanos field. For durations of one second or more, a non-zero value for the nanos field must be of the same sign as the seconds field. Must be from -999,999,999 to +999,999,999 inclusive.

Tool Annotations

Destructive Hint: ❌ | Idempotent Hint: ✅ | Read Only Hint: ✅ | Open World Hint: ❌