Prometheus Metrics
mcponce provides native, zero-dependency Prometheus metrics export compliant with the standard Prometheus text format (text/plain; version=0.0.4). You can scrape your MCP servers directly with Prometheus, VictoriaMetrics, Datadog, or Grafana Agent without needing external sidecar exporters.
The /metrics Endpoint
#
Every mcponce HTTP server exposes /metrics by default:
curl http://localhost:3000/metricsExample Prometheus output:
# HELP mcp_server_uptime_seconds Total time the server has been running in seconds.
# TYPE mcp_server_uptime_seconds gauge
mcp_server_uptime_seconds 3600.25
# HELP mcp_active_sessions Current active MCP client connections.
# TYPE mcp_active_sessions gauge
mcp_active_sessions 2
# HELP mcp_invocations_total Total number of tool invocations partitioned by tool and status.
# TYPE mcp_invocations_total counter
mcp_invocations_total{tool="get_weather",status="success"} 450
mcp_invocations_total{tool="get_weather",status="failure"} 3
mcp_invocations_total{tool="get_weather",status="cached"} 120
# HELP mcp_tool_retries_total Total retry attempts executed for tools.
# TYPE mcp_tool_retries_total counter
mcp_tool_retries_total{tool="get_weather"} 5
# HELP mcp_invocation_duration_seconds_avg Average duration of tool executions in seconds.
# TYPE mcp_invocation_duration_seconds_avg gauge
mcp_invocation_duration_seconds_avg{tool="get_weather"} 0.0452
# HELP mcp_cache_size Current number of items in memory cache.
# TYPE mcp_cache_size gauge
mcp_cache_size 85
# HELP mcp_cache_hits_total Total number of cache hits.
# TYPE mcp_cache_hits_total counter
mcp_cache_hits_total 120Programmatic Metric Export #
If you embed mcponce inside custom HTTP routers (Express, Fastify, Next.js, or Cloudflare Workers), generate Prometheus output directly with app.getMetrics():
import { createMcpServer } from 'mcponce';
const app = createMcpServer({ name: 'my-service' });
// Export formatted string
const prometheusText = app.getMetrics({
serverName: 'production-cluster-01',
version: '1.2.0'
});
console.log(prometheusText);Prometheus Scrape Configuration #
Add your mcponce server to your prometheus.yml:
scrape_configs:
- job_name: 'mcponce_servers'
scrape_interval: 15s
static_configs:
- targets: ['localhost:3000']Recommended Alert Rules #
| Metric Alert | Threshold | Action |
|---|---|---|
| High Error Rate | rate(mcp_invocations_total{status="failure"}[5m]) > 0.05 | Check upstream API keys or network status. |
| Excessive Retries | rate(mcp_tool_retries_total[5m]) > 2 | Third-party service is experiencing flakiness or throttling. |
| Spike in Duration | mcp_invocation_duration_seconds_avg > 5.0 | Investigate database query plans or network latency. |