What problem does it solve?
Creates and iterates Grafana dashboards without guessing Prometheus metric names, while producing consistent, import-safe dashboard JSON and a Git-ready ConfigMap.
Core Features & Use Cases
- MCP-driven metric discovery: Lists metric names, inspects metric metadata, explores label values, and tests queries before building panels.
- Dashboard JSON authoring with guardrails: Enforces required Grafana JSON conventions (e.g.,
id: null, graphTooltip: 1, hardcoded Prometheus datasource uid).
- GitOps persistence via ConfigMap sidecar compatibility: Produces a ConfigMap with the required
grafana_dashboard: "true" label and correct |- JSON body, plus registration guidance for kubernetes/.../kustomization.yaml.
- Use Cases: Creating new dashboards, modifying existing dashboard JSON, adding panels/sections, debugging blank/broken panels, and laying out dashboard grids.
Quick Start
Use the grafana-dashboards skill to author a new dashboard by first discovering the required Prometheus metrics via MCP, then pushing a draft to Grafana for screenshot review, and finally writing the Git-ready ConfigMap for the monitoring sidecar.