Google Cloud Monitoring Metric Finder
Discover and filter relevant monitoring metrics for GCP services.
The skill restricts itself to metric-descriptor retrieval and web search, mainly reading monitoring metadata, with no malware, credential theft, or covert exfiltration evident; however, it instructs the agent to directly modify local MCP configuration and use Google credentials when a tool is missing, without user confirmation, data-flow disclosure, explicit permission boundaries, or rollback steps, so substantial points are deducted.
The workflow covers pagination, filtering, timeout, rate-limit, and empty-result handling and specifies user feedback after failures; however, its claimed configuration-editing behavior is not present in allowed-tools, it depends on a particular MCP tool and environment paths, and no tests or execution evidence are supplied in this static review, warranting only a moderate score.
The scenario is clear: discover GCP metric descriptors, filter by keywords, and align them with monitored resources, with several prefix mappings documented; trigger and non-fit boundaries, required project permissions, and Chinese-language output are not sufficiently specified. The core function also depends on access to Google Cloud Monitoring MCP services, which may be unreachable in some mainland-China network environments, so points are deducted.
The name, description, staged workflow, tool dependency, troubleshooting guidance, output schema, and reference documentation are reasonably clear. The repository supplies Apache-2.0 licensing, installation guidance, official organization provenance, and a support path. However, the skill lacks its own version, changelog, named maintenance responsibility, update path, FAQs, examples beyond inline snippets, and explicit known-limitations disclosure.
The goal and required Markdown table fields are explicit, and results could be directly usable if the MCP returns complete descriptors. However, retrieving all descriptors and filtering them in context creates cost and quality dependence on keyword mappings; there is no verified representative output, ranking policy, or ambiguity handling. The score is conservatively limited by the static calibration.
The skill links to Google Cloud documentation and an MCP reference, and its payload, pagination, and field-mapping rules are auditable; however, it provides no committed tests, CI coverage, execution logs, or independent reproduction evidence. Key claims therefore remain primarily documentation assertions, justifying a limited static score.
- When list_metric_descriptors is unavailable, the skill directs the agent to write local MCP configuration and enable Google credentials; explicit user consent, exact-file disclosure, permission scope, and rollback should be required.
- The declared allowed-tools do not include filesystem or MCP-configuration write capabilities, so the auto-configuration procedure may be unrunnable.
- The mandatory Markdown-table output may conflict with callers that require JSON, CSV, or another format.
- The skill should clarify project-ID and credential requirements, potentially sensitive monitoring metadata, and the data flow to the external MCP service.
- No test or CI evidence is provided, so static review cannot confirm the key paths for pagination, filtering, resource alignment, or failure handling.
What does this skill do, and when should you use it?
This skill retrieves Google Cloud Monitoring metric descriptors for a GCP service or resource. It must query the live Google Cloud Monitoring API through the list_metric_descriptors MCP tool, then filter the results locally in the agent context. It examines metric types, names, descriptions, labels, and monitored resource types. It is suited to users who need to identify metric descriptors and schemas, but it depends on an available MCP service and network access.
It identifies the target GCP service, project, and metric concepts from the request; calls list_metric_descriptors with service-specific filters and follows pagination tokens; matches keywords against each descriptor's type, display name, and description; checks labels against the requested resource granularity; and returns 5 to 15 relevant metrics per service in Markdown tables containing metric type, display name, description, metric kind, value type, and monitored resource types.
- A GCP operator needs CPU or memory metrics for Compute Engine.
- A monitoring engineer must find metric types and value schemas for BigQuery, Cloud Run, or Cloud SQL.
- A platform team needs metrics whose labels align with a particular monitored-resource granularity.
- A developer wants to confirm the descriptor, metric kind, and value type for a GCP service metric.
- An incident responder needs a defined retry and documentation-checking path when metric discovery fails or returns no results.
What are this skill's strengths and limitations?
- Requires live descriptor retrieval, which helps account for changing GCP metric definitions.
- Retrieves all available descriptors through pagination before local filtering.
- Checks metric labels and monitored resource types for resource alignment.
- Includes retry and public-documentation fallback procedures for failed queries.
- Requires the Google Cloud Monitoring MCP tool, network access, and project availability.
- It selects metric descriptors rather than creating alerts or analyzing time-series data.
- Keyword matching works best when the request states clear metric concepts.
- The source provides no test suite, performance measurements, or broad client compatibility verification.
How do you install this skill?
In an Agent Skills-compatible environment, run: npx skills add google/skills. Select the cloud-monitoring-metric-selection skill during installation. The source does not document a client-specific installation directory or additional setup; use also requires an available Google Cloud Monitoring MCP service.
How do you use this skill?
After installation, make a natural-language request such as: "Find the relevant Cloud Run latency and request-count metrics for project my-project." The skill extracts the service and metric keywords, queries Google Cloud Monitoring live, and returns a service-grouped Markdown table. The source does not define a more specific trigger command or client syntax.