Skip to content

grafana-mcp-prompts.md

Latest commit

 

History

History
54 lines (38 loc) · 1.69 KB

grafana-mcp-prompts.md

File metadata and controls

54 lines (38 loc) · 1.69 KB
Raw

Grafana MCP Tool - Example Prompts

Infrastructure Overview

  • Application: Node.js Fastify API with OpenTelemetry instrumentation
  • Datasources: Prometheus (metrics), Loki (logs), Tempo (traces)
  • Services: PostgreSQL
  • Monitoring: Blackbox Exporter for health checks, OpenTelemetry Collector

📊 Query HTTP Request Metrics

Prompt:

Query Prometheus for HTTP request rate and latency metrics from the alumnus application over the last hour. Show me the request rate per endpoint and the p95 latency.

What it does:

  • Queries http_server_request_duration_seconds metrics
  • Shows request rates and latency percentiles
  • Helps identify slow endpoints and traffic patterns

🔥 Check Firing Alerts

Prompt:

List all currently firing alerts from Prometheus and show me their severity, description, and how long they've been firing.

What it does:

  • Lists alerts in "firing" state
  • Shows critical infrastructure issues
  • Provides alert metadata and duration

Service Dependency Map

Query Tempo to generate a service dependency map showing all services called during a specific trace. Visualize the request flow through the microservices architecture.

Tips for Effective Queries

  1. Time Ranges: Always specify time ranges (e.g., "last hour", "last 30 minutes") for focused results
  2. Label Filters: Use labels to filter metrics by service, instance, or job name
  3. Aggregations: Use rate(), avg(), max() functions in PromQL for meaningful metrics
  4. Trace Correlation: Leverage trace IDs in logs to jump between Loki and Tempo
  5. Alert Context: When checking alerts, ask for severity, duration, and affected services