prometheus-mcp-server
# prometheus-mcp-server
A Model Context Protocol (MCP) server for interacting with Prometheus metrics and data.
TypeScript-based MCP server implementing a Prometheus API interface. Bridges Claude and your Prometheus server via MCP.
## Features
- Instant and range PromQL queries
- Series and label discovery
- Metric metadata access
- Scrape target info
- Alerts, rules, and status endpoints
## Installation
```bash
npm install -g prometheus-mcp-server
# or
npm install prometheus-mcp-server
# or
npx prometheus-mcp-server
```
## Usage
```bash
npm install
npm run build
npm start
```
Set the `PROMETHEUS_HOST` environment variable to your Prometheus instance URL.
Example Claude Desktop config:
```json
{
"mcpServers": {
"prometheus-mcp-server": {
"command": "npx prometheus-mcp-server",
"env": {
"PROMETHEUS_HOST": "http://your-prometheus-instance:9090"
}
}
}
}
```
## Functions
- `mcp__instant_query`
- `mcp__range_query`
- `mcp__get_series`
- `mcp__get_label_values`
- `mcp__get_metadata`
- `mcp__get_targets`
- `mcp__get_alerts`
- `mcp__get_rules`
- `mcp__get_status`
TDQS
Scored across 10 tools
Every tool has a clearly distinct purpose targeting specific Prometheus resources or query types, with no overlapping functionality. For example, get-alerts retrieves alerts, get-series finds series by labels, and instant-query versus range-query handle different query modes, making misselection unlikely.
The naming follows a consistent verb_noun pattern with 'get-' or query-type prefixes for most tools, making them predictable. However, 'context-prometheus' deviates slightly by using a noun_verb structure and being in Portuguese, which breaks the pattern but is still understandable.
With 10 tools, this is well-scoped for a Prometheus server, covering key operations like queries, metadata, alerts, rules, and status without being overwhelming. Each tool serves a distinct and necessary function in monitoring and data retrieval.
The toolset provides comprehensive coverage for querying, monitoring, and status checks in Prometheus, including instant and range queries, alerts, rules, and targets. A minor gap exists in management operations like configuration changes or data deletion, but core workflows are fully supported.