Grafana MCP
by Grafana Labs
AI-powered access to Grafana dashboards, alerting, logs, and incident management
devops Python Intermediate Self-hostable Docker Verified
⭐ 2.4k stars 📅 Updated: 1mo ago
Description
MCP server from Grafana Labs for AI assistants to access Grafana instances. Comprehensive coverage of the Grafana ecosystem including dashboard operations, datasource management (Prometheus, Loki, ClickHouse, CloudWatch, Elasticsearch), PromQL query execution, alert rule management, log search, incident management, on-call scheduling, and panel image rendering. Multiple installation methods including uvx, Docker, binary, and Helm chart for Kubernetes. Built in Go for high performance and minimal resource usage. Ideal for DevOps and SRE teams wanting AI-assisted observability workflows.
✅ Best for
DevOps and SRE teams using Grafana who want AI-assisted monitoring, alerting, and incident response
⏭️ Skip if
You don't use Grafana for observability or have a different monitoring stack
💡 Use cases
- Searching and exploring Grafana dashboards using natural language queries
- Executing PromQL queries and analyzing Prometheus metrics from AI assistants
- Managing alert rules and investigating incidents through conversational AI
- Searching logs across Loki, Elasticsearch, and other datasources
👍 Pros
- ✓ Official Grafana Labs project with comprehensive ecosystem coverage
- ✓ Multiple install methods — uvx, Docker, binary, Helm for Kubernetes
- ✓ Supports all major datasources — Prometheus, Loki, ClickHouse, CloudWatch, Elasticsearch
- ✓ Built in Go for excellent performance and low resource footprint
👎 Cons
- ✗ Requires a running Grafana instance with service account configured
- ✗ Complex permission model — service account needs appropriate access levels
- ✗ Some advanced features require Grafana Cloud or Enterprise edition
🔧 Exposed tools (8 tools)
| Tool | Category | Description |
|---|---|---|
| manage_alerts | alerting | Create, update, and manage Grafana alert rules |
| search_dashboards | dashboards | Search for Grafana dashboards by name or tag |
| get_dashboard | dashboards | Retrieve a specific dashboard by UID with full panel details |
| manage_incidents | incidents | Create and manage incidents in Grafana Incident |
| search_logs | logs | Search logs across Loki, Elasticsearch, and other log datasources |
| manage_oncall | oncall | Manage on-call schedules and escalation policies |
| query_prometheus | query | Execute PromQL queries against Prometheus datasources |
| render_panel | rendering | Render a dashboard panel as an image |
⚡ Installation
Prerequisites:
- • python v3.10+
- • Grafana service account token
- • API key required
Check Claude Code documentation to configure this MCP server.
💡 Tips & tricks
Create a dedicated Grafana service account with read-only permissions for safe exploration. Use the Helm chart for Kubernetes-native deployments. The query_prometheus tool is especially powerful for ad-hoc metric analysis during incident investigation.
Quick info
- Author
- Grafana Labs
- License
- Apache-2.0
- Runtime
- Python 3.10+
- Transport
- stdio
- Category
- devops
- Difficulty
- Intermediate
- Self-hostable
- ✅
- Auth
- —
- Docker
- 🐳 Docker available
- Version
- latest
- Updated
- Feb 28, 2026
Client compatibility
- ✅ Claude Code
- ✅ Cursor
- ❓ VS Code Copilot
- ❓ Gemini CLI
- ❓ Windsurf
- ❓ Cline
- ❓ JetBrains AI
- ❓ Warp
Platforms
🍎 macOS 🐧 Linux 🪟 Windows