Prometheus MCP Server

I was unable to find a Model Context Protocol (MCP) server implementation for Prometheus. I had a quiet weekend and so I’ve been writing one: prometheus-mcp-server.

I used the code from the MCP for gRPC Health Checking protocol that I wrote about previously as a guide.

I wrote a series of stdin and HTTP tests to have confidence that the service is working correctly but I had no MCP host.

I discovered that Visual Studio Code through its GitHub Copilot extension functions has a preview to use MCP servers i.e. function as an MCP host and access MCP servers.

And, it works just great!

1. Ensure you have the GitHub Copilot extension installed.
2. Run prometheus-mcp-server (assuming on default http://localhost:7777)
3. Add the prometheus-mcp-server either by CTRL-SHIFT-P “MCP: Add Server” or manually adding .vscode/mcp.json (see below)
4. Configure GitHub Copilot to use “Agent” mode (see below)
5. From the GitHub Copilot change click the tools icon (see below)

.vscode/mcp.json:

{
    "servers": {
        "prometheus-mcp-server": {
            "url": "http://localhost:7777/mcp"
        }
    }
}

NOTE The url must include the /mcp path

GitHub Copilot “Agent” mode

GitHub Copilot tools

Now, you should (!) be able to interact with prometheus-mcp-server from the GitHub Copilot chat window, e.g.

#ping

And:

Unfortunately metrics (which lists all metrics) isn’t working correctly:

#metrics

But:

But, it is possible to query the metrics:

#query up{job="prometheus"}

And:

So, I followed up (with the agent itself) about the fact that it ran the metrics tool but did not provide the results.

Here’s the interaction:

• go_gc_cycles_automatic_gc_cycles_total
• go_gc_cycles_forced_gc_cycles_total
• go_gc_cycles_total_gc_cycles_total
• …

If you want details or values for any of these, let me know!

1. Display the number of results found (e.g., “There are 120 metrics available.”).
2. Prompt the user to provide a filter (e.g., “Please specify a keyword or prefix to filter the results, such as go_, process_, or prometheus_.”).

This approach helps avoid overwhelming output and guides the user to get relevant information efficiently. If you want, I can help you implement this logic in your code or workflow. Let me know your preferred language or tool!

What you can do:

• …
• If you want the Copilot agent to always behave this way, you would need to request this as a feature from the Copilot team or adjust your workflow to always ask for a filter.

Per the Agent’s suggestion, I filed an issue

That’s all (for now)!

• MCP
• Prometheus