Prometheus MCP server

Last verified: 2026-09

Instant queries against Prometheus.

Community Cloud & infra Needs a secret Claude Desktop, Cursor, Windsurf, Claude Code

What it does

PromQL against a Prometheus (or compatible) URL. Instant queries, range queries, metric discovery, targets, health.

It sits in Cloud & infra: Most of these are admin APIs.

Good for

execute_query with up or a PromQL you already trust.
execute_range_query on a short window — huge ranges time out.
list_metrics / get_targets when you do not remember the name.
Point PROMETHEUS_URL at a read endpoint, not an admin proxy.

Tools

health_check — Is the server up.
execute_query / execute_range_query — PromQL.
list_metrics / get_targets — Discovery.

Config (Claude Desktop / Cursor / Windsurf)

Paste the JSON below. Same mcpServers shape. Replace placeholder paths and secrets. Windows paths look like C:\\Users\\you\\project, not /path/to. Host-side steps →

Claude Desktop

OS	Config file
macOS	`~/Library/Application Support/Claude/claude_desktop_config.json`
Windows	`%APPDATA%\Claude\claude_desktop_config.json` (usually `C:\Users\<you>\AppData\Roaming\Claude\`)
Linux	`~/.config/Claude/claude_desktop_config.json`

Cursor

Scope	macOS / Linux	Windows
This project	`.cursor/mcp.json` in the repo root
This user	`~/.cursor/mcp.json`	`%USERPROFILE%\.cursor\mcp.json`

Windsurf

OS	Config file
macOS / Linux	`~/.codeium/windsurf/mcp_config.json`
Windows	`%USERPROFILE%\.codeium\windsurf\mcp_config.json`

{
  "mcpServers": {
    "prometheus": {
      "command": "npx",
      "args": [
        "-y",
        "prometheus-mcp-server"
      ],
      "env": {
        "PROMETHEUS_URL": "http://localhost:9090"
      }
    }
  }
}

Windsurf remote MCP: use serverUrl instead of url if the block below is HTTP.

One-liner: npx -y prometheus-mcp-server

Secrets it wants: PROMETHEUS_URL

Source repo →

How to get started

Point PROMETHEUS_URL at a read endpoint. Add token/basic auth if the instance needs it.
Run a small instant query you already know (up).
Range queries over huge windows will time out and load the server.

Access risk

Usually read-only. Still: query load on Prometheus and any secrets sitting in metric labels.

When to skip it

Skip it if Grafana already proxies the datasource, or if the metrics live only in Datadog.

Instead: Grafana, Datadog, Elasticsearch.

Cloud & infraGrafanaDashboards and datasources. Cloud & infraDatadogMetrics, logs, and monitors. Databases & dataElasticsearchSearch an Elasticsearch cluster.

Vs alternatives

Server	Official?	Needs a secret?	Best for
Prometheus	No	Yes	Instant queries against Prometheus.
Grafana	No	Yes	Dashboards and datasources.
Datadog	No	Yes	Metrics, logs, and monitors.
Elasticsearch	No	Yes	Search an Elasticsearch cluster.

More in Cloud & infra

OfficialAWS Knowledge BaseQuery an Amazon Bedrock knowledge base. Cloud & infraAWSCall AWS APIs with the default credential chain. Cloud & infraAzureAzure Resource Manager via MCP. Cloud & infraCloudflareWorkers, KV, R2, and DNS via Cloudflare.

All 15 in this category →

FAQ

Does it write?

Usually read-only PromQL. Risk is query load and secrets sitting in metric labels.

Prometheus or Grafana?

This talks to Prometheus (or compatible) directly. Grafana if you want dashboards and mixed datasources.

Auth?

Optional token/basic/client cert — see the repo. PROMETHEUS_URL is the minimum.

Can a query take down Prometheus?

Wide range queries can. Start with instant queries you already run.