MetricFire MCP Server

official

Monitor your infrastructure using natural language with the new MetricFire Model Context Protocol (MCP) Server. Connect the MetricFire MCP Server to AI assistants like GitHub Copilot and use simple prompts to search metrics, visualize data, create alerts, and manage your monitoring environment, without writing API calls.

What can you do with MetricFire MCP?

  • Search metrics — Ask your assistant to find metrics by pattern or tags, including busy or invalid metrics.
  • Render Graphite data — Request metric data as JSON, CSV, raw, PNG, or SVG for a specified time range.
  • Manage alerts — Create, update, search, mute, or delete alerts, including composite alerts with AND/OR logic.
  • Build Grafana dashboards — Have your assistant create or update dashboards and panels from natural-language prompts.
  • Handle notification channels — Retrieve, create, update, or delete channels like Email, Slack, PagerDuty, and webhooks.
  • Schedule mutes — Set up recurring mute windows (e.g., weekends) and associate them with alerts.

Documentation

For the complete documentation index, see llms.txt. Markdown versions of documentation pages are available by appending .md to page URLs; this page is available as Markdown.

MF MCP Server

Connect AI assistants to interact with Hosted Graphite API's using natural-language prompts.

Contents

{% embed url="https://www.youtube.com/watch?t=15s&v=7dIbd2rxARs" %}

The MetricFire MCP Server allows supported AI clients to interact directly with your Hosted Graphite account using the Model Context Protocol (MCP).

Once connected, clients such as GitHub Copilot, Cursor, Claude/Claude Code, etc. can use Hosted Graphite APIs for metric discovery, Graphite rendering, alert management, notification channels, scheduled mutes, and other monitoring workflows.

The MCP server operates on data and resources in your Hosted Graphite account. Metrics must already be forwarded and stored in Hosted Graphite before they can be searched, rendered, or used for alerting.

Create an MCP Token

In your Hosted Graphite account, navigate to Add-Ons ⇒ MetricFire MCP Server

Create a token and select the permissions required by your MCP client:

  • Read – access metrics, renders, alerts, notification channels, scheduled mutes, and other supported resources.
  • Write – create or modify supported resources.
  • Delete – delete supported resources.

We recommend using the minimum permissions required for your workflow.


The full MCP token is only displayed once when it is created. Store it securely before leaving the page.

Connect an MCP Client

The Hosted Graphite MCP endpoint is:

https://mcp.hostedgraphite.com/

Configuration varies slightly depending on the MCP client.

GitHub Copilot (VS Code)

Open the VS Code Command Palette and select: MCP: Open User Configuration

Add the MetricFire MCP Server object:

{
  "servers": {
    "mf-mcp-server": {
      "url": "https://mcp.hostedgraphite.com/",
      "headers": {
        "X-HostedGraphite-MCP-Token": "YOUR-MCP-TOKEN"
      }
    }
  }
}

Save the configuration, start the MCP server from within VS Code, and interact with the agent using Copilot Chat.

Cursor, Claude Code, Cline, Windsurf

These clients support custom HTTP headers. Configure the endpoint:

https://mcp.hostedgraphite.com/

and pass your token:

X-HostedGraphite-MCP-Token: <MF-MCP-TOKEN>

Refer to your client's MCP documentation for the location and format of its server configuration.

Claude

Claude's browser/UI configuration can connect using the MCP token in the server URL:

https://<MF-MCP-TOKEN>@mcp.hostedgraphite.com

NOTE: When using URL-based authentication, we highly recommend a read-only token unless write or delete access is specifically required.

Available Tools

After connecting, your MCP client can discover the tools currently exposed by Hosted Graphite.

Metrics

Tools are available for:

  • Standard metrics search (by pattern*)
  • Tagged metrics search
  • Busy metrics
  • Invalid metrics
  • Metric deletion

Graphite Render

The Graphite Render API can be queried through MCP, with support for:

  • JSON
  • CSV
  • Raw data
  • PNG
  • SVG

This allows MCP clients to locate metrics and retrieve or visualize their data over a specified time range.

Alerts

AI assistants can work with your Hosted Graphite metrics to build meaningful composite alerts using AND/OR logic, making it easier to alert on service-level issues rather than individual metric thresholds.

For example: “Find the relevant CPU, memory, disk, and network metrics for my production servers and create a composite alert that emails me when multiple signals indicate a service-level issue.”

Alerting tools support:

  • Creating and updating alerts
  • Searching and retrieving alerts
  • Deleting alerts
  • Muting and unmuting alerts
  • Alert status and history
  • Composite alert criteria using AND/OR logic

Dashboards

AI clients can reference metrics stored in your Hosted Graphite account and use them to build, update, and manage Grafana dashboards using natural-language prompts. Instead of manually searching for metrics, creating panels, and configuring visualizations one at a time, you can ask your AI assistant to build a dashboard around a service, host, application, or infrastructure component you want to monitor.

For example: “Find the relevant CPU, memory, disk, and network metrics for my production servers and build a comprehensive infrastructure overview dashboard.”

Dashboard tools support:

  • Creating Grafana dashboards
  • Updating existing dashboards
  • Searching and retrieving dashboards
  • Deleting dashboards
  • Creating and configuring dashboard panels
  • Using Hosted Graphite metrics to build dashboards from natural-language prompts

{% embed url="https://www.youtube.com/watch?v=jzkK_87fi0Q" %}

Notification Channels

MCP clients can work with Hosted Graphite notification channels (Email, Slack, PagerDuty, MS Teams, webhooks, etc.), including retrieving channels and associating them with alerts. Depending on token permissions, supported channels can also be created, updated, or deleted.

Scheduled Mutes

Scheduled mute tools allow MCP clients to create and manage recurring mute windows (e.g; weekends). Scheduled mutes are separate resources that can be associated with alerts through the Alert API.

How It Works

The MetricFire MCP Server provides an MCP interface to supported Hosted Graphite APIs. The connected AI client selects the appropriate MCP tool based on the request and executes it using the permissions associated with your MCP token.

The MCP integration does not replace your existing metric collection. Continue sending data to Hosted Graphite using your existing agents, integrations, or supported protocols.

For more details about the underlying APIs, see the Hosted Graphite API documentation and if you need help connecting an MCP client, contact us at: support@metricfire.com

MCP is not currently available for users that have enable the Hosted Graphite Add-on from the Heroku Marketplace. Contact us to learn how to migrate your Heroku HG account(s) to a dedicated HG account and take advantage of all available features like MCP and Logging.


Agent Instructions

This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the ask and goal query parameters:

GET https://docs.hostedgraphite.com/add-ons-and-integrations-guide/mf-mcp-server.md?ask=<question>&goal=<user_goal>

ask is the immediate question: it should be specific, self-contained, and written in natural language. goal is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with ask=how do I create an API token, a goal like automate deployments from our CI pipeline lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.