Context4
Context4MCP connector / BETA

Grafana MCP connector

Find dashboards and examine Grafana metrics and logs through AI.

Start with your task

Grafana — in your AI.
Start with one task.

Copy the prompt for your AI. You will confirm the account connection and access permission separately.

What you can ask AI to do ↓

Your data. Your tasks.

What to ask AI to do in Grafana

Start with one clear task. These examples use the current connector's tools; results depend on your account data and permissions.

01

Find a dashboard

«Find Grafana dashboards by my service name.»

AI will request matching dashboards and help choose the right one.

02

Review metrics

«Run this Prometheus query in Grafana for the selected period and help me understand the result.»

AI will retrieve time series and can discuss them with you.

03

Examine logs

«Run this Loki query through Grafana and help me examine the logs.»

AI will request available logs within the specified parameters.

From prompt to first task

How to connect Grafana to AI

  1. 01

    Copy the prompt

    Choose an AI client in the setup block. Paste the prompt into your chat and send it: it explains how to configure Context4 and find the service you need.

  2. 02

    Connect your account

    Sign in to Context4 and add your Grafana API key through the secure connection flow. Do not send the key in a regular chat.

  3. 03

    Assign your first task

    Start by reading data from Grafana. AI will check the available tools and connection, then perform a suitable request if possible.

For ChatGPT, you need to add an HTTPS-accessible MCP server manually. Connection availability depends on your account and workspace settings. ChatGPT guide →

Your account.
Your permissions.

Context4 gives AI tools to work with Grafana. Data and actions are available within the connected account's permissions.

AI may receive an empty response or an access error. To make changes, it needs an appropriate tool, parameters and your instruction.

How connections and authorization work →Context4: Grafana MCP connector, an AI request and available actions

Before you connect

Questions about Grafana and MCP

What does the Grafana MCP connector provide?

It gives your AI tools for interacting with Grafana. Find dashboards and examine Grafana metrics and logs through AI. Available actions depend on the connector implementation and the connected account's permissions. Learn more about how MCP works in Context4.

What tasks can I assign to AI in Grafana?

For example: find a dashboard, review metrics, examine logs. Example requests appear above. The answer is based on data returned by the service; missing data or insufficient permissions may prevent a result.

What do I need to connect Grafana?

An account in Grafana, permission to access the required data and an AI client that supports MCP. Grafana uses an API key: prepare it in the service settings and add it through the secure Context4 connection flow. Details are in our service connection guide.

Can I connect Grafana to my AI?

The setup guide has separate paths for ChatGPT, Claude, Codex, Claude Code, Hermes and OpenClaw. ChatGPT needs a plugin; Codex can connect MCP directly. The prompt explains the steps, but does not automatically install a missing plugin. Access depends on client capabilities and your account permissions. End-to-end compatibility of Hermes and OpenClaw with Context4 has not yet been verified.

Can AI change data in Grafana?

Only if the connector provides the relevant tool and it is available to your connection. Explicitly request creation, changes or sending: specify the object and required action. Check the result afterwards. Connectors do not all support the same operations.

How do I disconnect access to Grafana?

Disconnect the account in Context4 and revoke the permission or key in the settings of Grafana. Do not paste passwords or secret keys into a regular AI chat. Learn more about access management.

How much does using Grafana through Context4 cost?

Context4 is free. See the terms of use or support. Subscriptions to Grafana and the AI client may be billed separately under each service's rules.

Technical reference Grafana

Version 1.2.0 · beta · 38 actions in the catalog. Permissions and availability are checked when connecting.

MCP connector actions

  • Read configured alert state history

    alert_history.query

  • List contact points

    alert.contact_points

  • Get an alert rule definition

    alert.rule

  • List alert rules with their state

    alert.rules

  • Create a silence

    alert.silence

  • List silences

    alert.silences

  • Read annotations in a UTC window

    annotation_range.list

  • Create an annotation

    annotation.create

  • List annotations

    annotation.list

  • Resolve dashboard incident queries

    dashboard_query.resolve

  • Get a dashboard's full model

    dashboard.get

  • Get a dashboard's summary with recent versions

    dashboard.overview

  • List a dashboard's panel queries

    dashboard.panel_queries

  • Save a dashboard

    dashboard.save

  • Search dashboards

    dashboard.search

  • Summarise a dashboard

    dashboard.summary

  • List a dashboard's versions

    dashboard.versions

  • Query datasource data frames

    datasource_query.frames

  • Query an incident datasource

    datasource_query.run

  • Get a datasource

    datasource.get

  • Check a datasource

    datasource.health

  • List datasources

    datasource.list

  • Create a folder

    folder.create

  • Search folders

    folder.search

  • Check the Grafana instance

    instance.health

  • Overview of a service's errors in logs

    loki.error_overview

  • List Loki label values

    loki.label_values

  • List Loki label names

    loki.labels

  • Detect log patterns

    loki.patterns

  • Query Loki logs

    loki.query

  • Estimate a Loki query's size

    loki.stats

  • List Prometheus label values

    prometheus.label_values

  • List Prometheus label names

    prometheus.labels

  • Get Prometheus metric metadata

    prometheus.metadata

  • Run an instant PromQL query

    prometheus.query

  • Run a PromQL range query

    prometheus.query_range

  • Read a complete trace

    tempo.get

  • Search traces with TraceQL

    tempo.search

Data and authorization

Authorization: API key.

    No synchronized data types are claimed.