Quickstart

Install

uv add lexigram-monitor

For production monitoring, install your backend extra:

# Prometheus metrics
uv add "lexigram-monitor[prometheus]"
# OpenTelemetry tracing (OTLP)
uv add "lexigram-monitor[otel]"
# System metrics (CPU, memory)
uv add "lexigram-monitor[system]"
# All extras
uv add "lexigram-monitor[all]"

Minimal Example

import asyncio

from lexigram import Application, LexigramConfig
from lexigram.monitor import MonitorProvider
from lexigram.monitor.noop import NoOpMetricsBackend


async def main():
    config = LexigramConfig.from_yaml("application.yaml")
    app = Application(name="my-app", config=config)
    app.add_provider(MonitorProvider(backend=NoOpMetricsBackend()))

    async with app.boot():
        # Create and increment a counter
        provider = await app.container.resolve(MonitorProvider)
        counter = provider.create_counter("requests_total", "Total requests")
        counter.increment()
        counter.increment(5)
        print(f"Count: {counter.get_count()}")

        # Record an HTTP request
        provider.record_request("GET", "/api/users", 0.125, 200)

        # Read health
        health = await provider.health_check()
        print(f"Status: {health.status}")


asyncio.run(main())

Wiring

Add MonitorProvider to your Application. The provider auto-discovers MonitorConfig from the monitor: section of your YAML config:

from lexigram.monitor import MonitorProvider
from lexigram.monitor.config import MonitorConfig

# In-memory (development / testing)
app.add_provider(MonitorProvider(backend=NoOpMetricsBackend()))

# From config — reads LEX_MONITOR__* env vars
provider = MonitorProvider.from_config(MonitorConfig())
app.add_provider(provider)

Or use the declarative MonitorModule:

from lexigram.monitor import MonitorModule
from lexigram.monitor.backends.prometheus import PrometheusBackend

app.add_module(MonitorModule.configure(backend=PrometheusBackend()))

What Just Happened

MonitorProvider registered singletons for MetricsCollectorProtocol, TracerProtocol, and HealthCheckRegistryProtocol
The @traced and @metered decorators now resolve real implementations from the container
health_check() returns component status with backend type and metric count

Next Steps

Guide — mental model, core concepts, best practices
Configuration — every config key with defaults and env vars
How-Tos — common recipes