Add MCP usage guide for Grafana (#282)

Signed-off-by: Dan Barr <[email protected]>
Co-authored-by: Dan Barr <[email protected]>

Author: danbarr

docs/toolhive/guides-mcp/grafana.mdx

@@ -0,0 +1,229 @@
+---
+title: Grafana MCP server guide
+sidebar_label: Grafana
+description:
+  Using the Grafana MCP server with ToolHive for dashboard management,
+  datasource queries, alerting, and incident response.
+last_update:
+  author: danbarr
+  date: 2025-11-06
+---
+
+## Overview
+
+The official [Grafana MCP server](https://github.com/grafana/mcp-grafana)
+provides comprehensive access to your Grafana instance and its surrounding
+ecosystem. With support for dashboards, datasource queries (Prometheus, Loki,
+Pyroscope), alerting, incident management, Grafana OnCall, and Sift
+investigations, this server enables AI agents to interact with your entire
+observability stack.
+
+The server works with both local Grafana instances and Grafana Cloud, making it
+ideal for tasks like troubleshooting production issues, analyzing metrics and
+logs, managing dashboards, and coordinating incident response.
+
+## Metadata
+
+<MCPMetadata name='grafana' />
+
+## Usage
+
+You'll need a Grafana service account token to authenticate with the Grafana
+API. The token must have permissions for the Grafana features you want to access
+(such as dashboards, datasources, or alerting). Refer to the
+[Grafana service account documentation](https://grafana.com/docs/grafana/latest/administration/service-accounts/)
+for details on creating tokens and configuring permissions.
+
+<Tabs groupId='mode' queryString='mode'>
+<TabItem value='ui' label='UI' default>
+
+Select the `grafana` MCP server in the ToolHive registry.
+
+In the **Secrets** section, add your Grafana service account token or select an
+existing secret that contains the token.
+
+In the **Environment Variables** section, configure the connection to your
+Grafana instance:
+
+- `GRAFANA_URL`: Your Grafana instance URL (for example, `http://localhost:3000`
+  for local instances or `https://myinstance.grafana.net` for Grafana Cloud)
+- `GRAFANA_ORG_ID` (optional): The numeric organization ID if your Grafana
+  instance has multiple organizations
+
+:::tip[Security tip]
+
+Enable outbound network filtering on the **Network Isolation** tab to restrict
+the server's network access. Update the allowed hosts to match your Grafana
+instance domain.
+
+:::
+
+</TabItem>
+<TabItem value='cli' label='CLI'>
+
+Create a secret containing your Grafana service account token:
+
+```bash
+thv secret set grafana-token
+```
+
+Run the server with your Grafana instance URL and the secret:
+
+```bash
+thv run \
+  -e GRAFANA_URL=http://localhost:3000 \
+  --secret grafana-token,target=GRAFANA_SERVICE_ACCOUNT_TOKEN \
+  grafana
+```
+
+For Grafana Cloud, use your cloud instance URL:
+
+```bash
+thv run \
+  -e GRAFANA_URL=https://myinstance.grafana.net \
+  --secret grafana-token,target=GRAFANA_SERVICE_ACCOUNT_TOKEN \
+  grafana
+```
+
+Enable [network isolation](../guides-cli/network-isolation.mdx) to restrict the
+server's network access. Create a permission profile with your Grafana instance
+domain:
+
+```json title="grafana-profile.json"
+{
+  "network": {
+    "outbound": {
+      "insecure_allow_all": false,
+      "allow_host": ["myinstance.grafana.net"],
+      "allow_port": [443]
+    }
+  }
+}
+```
+
+Then run with the custom profile:
+
+```bash
+thv run \
+  -e GRAFANA_URL=https://myinstance.grafana.net \
+  --secret grafana-token,target=GRAFANA_SERVICE_ACCOUNT_TOKEN \
+  --isolate-network --permission-profile grafana-profile.json \
+  grafana
+```
+
+If your Grafana instance has multiple organizations, add the `GRAFANA_ORG_ID`
+environment variable with the numeric organization ID (for example,
+`-e GRAFANA_ORG_ID=2`).
+
+:::tip[Debug mode]
+
+Add the `--` separator followed by `-debug` to enable detailed logging of HTTP
+requests and responses:
+
+```bash
+thv run \
+  -e GRAFANA_URL=http://localhost:3000 \
+  --secret grafana-token,target=GRAFANA_SERVICE_ACCOUNT_TOKEN \
+  -- -debug
+```
+
+:::
+
+</TabItem>
+<TabItem value='k8s' label='Kubernetes'>
+
+Create a Kubernetes secret containing your Grafana service account token:
+
+```bash
+kubectl -n toolhive-system create secret generic grafana-token \
+  --from-literal=token=<YOUR_TOKEN>
+```
+
+Create a Kubernetes manifest to deploy the Grafana MCP server:
+
+```yaml {14-17} title="grafana.yaml"
+apiVersion: toolhive.stacklok.dev/v1alpha1
+kind: MCPServer
+metadata:
+  name: grafana
+  namespace: toolhive-system
+spec:
+  image: docker.io/mcp/grafana:latest
+  transport: sse
+  mcpPort: 8000
+  proxyPort: 8080
+  env:
+    - name: GRAFANA_URL
+      value: 'http://localhost:3000'
+  secrets:
+    - name: grafana-token
+      key: token
+      targetEnvName: GRAFANA_SERVICE_ACCOUNT_TOKEN
+```
+
+Apply the manifest to your Kubernetes cluster:
+
+```bash
+kubectl apply -f grafana.yaml
+```
+
+For Grafana Cloud, update the `GRAFANA_URL` in the manifest:
+
+```yaml
+spec:
+  env:
+    - name: GRAFANA_URL
+      value: 'https://myinstance.grafana.net'
+```
+
+If your Grafana instance has multiple organizations, add the `GRAFANA_ORG_ID`
+environment variable with the numeric organization ID:
+
+```yaml
+spec:
+  env:
+    - name: GRAFANA_URL
+      value: 'http://localhost:3000'
+    - name: GRAFANA_ORG_ID
+      value: '2'
+```
+
+</TabItem>
+</Tabs>
+
+## Sample prompts
+
+Here are some sample prompts you can use to interact with the Grafana MCP
+server:
+
+- "Show me all dashboards related to Kubernetes monitoring"
+- "Query the Prometheus datasource for CPU usage over the last hour for the
+  `api-service` pod"
+- "Get the recent alerts that are currently firing"
+- "List all open incidents and show me details for the most recent one"
+- "Find error patterns in the logs from the `production` namespace using Loki"
+- "Who is currently on call for the backend team schedule?"
+- "Create a new incident titled 'High memory usage on production cluster' with
+  severity critical"
+- "Show me the panel queries from the 'API Performance' dashboard"
+- "Get label values for the `namespace` label from the Loki datasource"
+- "List all Sift investigations from the past week"
+
+## Recommended practices
+
+- Create service accounts with least-privilege permissions. Use fine-grained
+  RBAC scopes to limit access to only the datasources, dashboards, and features
+  required for your specific use case.
+- Regularly rotate service account tokens and update the secrets in ToolHive.
+- Enable network isolation to restrict the server's outbound network access to
+  your Grafana instance domain only.
+- For dashboards with large JSON configurations, use the `get_dashboard_summary`
+  or `get_dashboard_property` tools to minimize context window usage instead of
+  retrieving the full dashboard with `get_dashboard_by_uid`.
+- When working with multi-organization setups, always specify the
+  `GRAFANA_ORG_ID` to ensure operations target the correct organization.
+- Enable [telemetry](../guides-cli/telemetry-and-metrics.mdx) to monitor API
+  calls and track which Grafana resources are being accessed.
+- For production deployments, consider using the debug mode temporarily to
+  troubleshoot connection or permission issues, but disable it once everything
+  is working correctly.