Guide · Agent skill

How to use the Grafana Dashboards agent skill

Updated 2026-10-29 · written with AI from the project's own documentation and checked against our catalogue data

The grafana-dashboards skill gives an AI coding agent guidance and example snippets for designing Grafana dashboards that show system and application metrics. It is for engineers who run Prometheus and Grafana and want dashboard JSON, queries and alert drafts produced quickly. After this guide you will know how to install the skill, how to phrase requests so the agent uses it, and what to check before you import anything into a live Grafana server.

Grafana Dashboardswshobson/agents · MIT Download the skill

Quick start

1Download the skill. Download the grafana-dashboards ZIP file from this directory page.
2Unzip into a skills folder. Unzip it so the grafana-dashboards folder sits in ~/.claude/skills/ for personal use or in .claude/skills/ inside a project.
3Or upload it in the Claude apps. Open Settings, go to Capabilities, then Skills, and upload the skill.
4Ask for a dashboard in plain language. Describe the service, the metrics and the data source, for example Prometheus. Mention the panels you want.
5Review the output. Check metric names, thresholds, alert values and panel types against your own setup and Grafana version.
6Import or provision it. Load the JSON into Grafana yourself, for example through the UI or a provisioning file, and test it before production use.

What the grafana-dashboards skill makes the agent do

An agent skill is a folder containing a SKILL.md file. When a task matches the skill's description, the agent reads that file and follows its guidance. This one covers creating and managing production Grafana dashboards for real-time visualization of system and application metrics.

Once the skill is active, the agent can draft dashboards for applications and infrastructure. It follows a layout where critical numbers sit at the top, key trends sit in time series graphs, and detailed tables or heatmaps sit below. It also applies the RED method (rate, errors, duration) to services and the USE method (utilization, saturation, errors) to resources.

Dashboard JSON with panels, grid positions, tags and refresh settings
Prometheus queries such as rate, error ratio and histogram_quantile for latency
Panel types including stat, time series, table and heatmap
Template variables for namespace and service filters
Alert rule examples, Grafana provisioning files and Terraform or Ansible examples
Starter layouts for API, infrastructure, database and application dashboards

Install the grafana-dashboards skill

Download the ZIP file for the skill from this directory page. Unzip it so that the grafana-dashboards folder, including its SKILL.md file, sits inside your skills folder. For a personal install, use ~/.claude/skills/. For a single project, use the .claude/skills/ folder inside the project.

If you use the Claude apps instead of a local folder, open Settings, go to Capabilities, then Skills, and upload the skill there. Either way, start a new session afterwards so the agent picks up the skill.

Personal install: ~/.claude/skills/grafana-dashboards/
Project install: .claude/skills/grafana-dashboards/ inside your project
Claude apps: Settings → Capabilities → Skills, then upload the skill

Trigger the skill with natural requests

You do not need a special command. The agent decides to use the skill when your request matches its description, which covers building monitoring dashboards, visualizing metrics and creating operational observability interfaces. Name the thing you want to see and the data source if you have one.

Requests that name the service, the metrics and the goal produce the most useful output. Mention Prometheus, since the examples assume Prometheus query syntax.

Build a Grafana dashboard for our checkout API showing rate, errors and p95 latency from Prometheus.
Create a Kubernetes node dashboard with CPU, memory, disk I/O and network panels.
Draft a dashboard with namespace and service variables so I can filter by team.
Write a Grafana provisioning file that loads my dashboard JSON from /etc/grafana/dashboards.

Read and adapt the example output

The skill's examples use the RED and USE methods as the basis for panel choice. Error rate is written as a percentage of total requests, and latency uses histogram_quantile over bucketed request durations. The example variables use label_values queries against kube_pod_info and kube_service_info, so they assume kube-state-metrics labels are present.

Treat every threshold, alert value and time range as a starting point. The skill's example alert fires when the error rate is above 5 percent, and the example default time range is the last 6 hours. Your own traffic should decide the real numbers. The examples also use the older graph panel type and a legacy-style alert block, so check them against the Grafana version you run before importing.

Replace the metric names in the queries with the ones your exporters actually emit.
Set thresholds from your own baseline, not the example values.
Confirm the panel types and alert format work in your Grafana version.
Set units on each panel so axis labels are correct.

sum(rate(http_requests_total{namespace="$namespace", service=~"$service"}[5m]))

Safety notes: scripts and network access

This skill contains no scripts. It is written as guidance and example snippets. It does not run code by itself, and it does not connect to your Grafana server or call Grafana's API. The files contain no such code, so the agent's output is text and JSON that you review and import yourself.

Because the output is only text, nothing changes in your Grafana instance until you import or provision the files. Review the JSON before you load it, and test it in a non-production folder first. The skill also cannot verify queries against your traffic or against your Grafana version.

No scripts are included in the skill.
The skill does not use the network.
Nothing is applied to Grafana until you import or provision the output.
Test imported dashboards in a separate folder before production use.

Tips for better dashboards

The skill's best practices are easy to apply and make dashboards easier to read. Start from community templates for the stack you run, use consistent names for panels and variables, and group related metrics in rows. Add a description to each panel so the next person understands what it shows.

For teams that keep dashboards in source control, the skill includes examples for Terraform's grafana_dashboard resource and for copying JSON files with Ansible. You can ask the agent to write either form for your repository.

Use the same colors and naming across dashboards.
Use variables so one dashboard serves many services.
Check each panel with several time ranges before you trust it.
Keep related service and infrastructure views in the same folder.

apiVersion: 1

providers:
  - name: "default"
    orgId: 1
    folder: "General"
    type: file
    disableDeletion: false
    updateIntervalSeconds: 10
    allowUiUpdates: true
    options:
      path: /etc/grafana/dashboards

What people use it for

Drafting a service dashboard with request rate, error rate and p95 latency for an API team.
Starting a Kubernetes node or pod dashboard with namespace and service variables.
Sketching a database dashboard with queries per second, connection pool usage and replication lag.
Writing a Grafana provisioning file or Terraform resource to keep dashboards in source control.

Frequently asked questions

What is the grafana-dashboards agent skill?

It is a skill for AI agents that adds guidance and example snippets for designing Grafana dashboards. It covers layouts, RED and USE methods, Prometheus queries, template variables, provisioning files and alert rules. It is text and examples only.

How do I install the grafana-dashboards skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or .claude/skills/ inside a project. You can also upload it in the Claude apps under Settings, then Capabilities, then Skills.

Is the grafana-dashboards skill free and what is its licence?

The skill is listed under the MIT licence. Check the licence file in the source repository for the full terms before you reuse it.

Does the skill connect to my Grafana server?

No. The skill does not connect to Grafana and does not call its API. It produces JSON and text that you import or provision yourself.

What do I need to use it?

You need Grafana, since the generated JSON is meant to be imported or provisioned into it. You also need Prometheus or a compatible metrics data source with the metric names used in the queries.

Can I use it with data sources other than Prometheus?

The examples assume Prometheus query syntax, so they are not a good fit for other data sources without rewriting the queries. Ask the agent to adapt the queries, and check the results yourself.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use this skill if you run Prometheus and Grafana and want well-structured dashboard JSON, queries and provisioning files drafted quickly, especially if you are new to observability. Pick something else if you need the agent to connect live to your Grafana instance, if you need non-Prometheus data sources without rewriting queries, or if you plan to page people on the example thresholds unchanged.

Download the skill