sentry-logger 0.1.0

Push logs from your Python app to the Sentry dashboard — AI-powered service health monitoring

Sentry Logger SDK (Python)

Push logs from your Python app to the Sentry dashboard.

Installation

pip install sentry-logger
# or from the repo:
pip install ./sdk/python

Quick Start — CLI + Browser OAuth

sentry-logger init --app-name "my-service" --dsn "http://localhost:9000"

What happens:

1. CLI requests a device session from the backend
2. Browser opens your sign-in page
3. Sign in with Google — your app is registered and an API key is created
4. CLI polls and stores credentials in ~/.sentry_logger/config.json

Check the currently linked app:

sentry-logger status

SDK Usage

from sentry_logger import init

# Reads api_key and dsn from ~/.sentry_logger/config.json
init()

import logging
logging.info("Hello from my app")
logging.error("Something went wrong")

Explicit configuration:

from sentry_logger import init

init(
    api_key="sk_...",
    dsn="http://localhost:9000",   # your backend URL
)

Environment Variables

Variable	Default	Description
LOGSENTRY_URL	http://localhost:9000	Backend ingest URL base

FastAPI Example

import logging
from fastapi import FastAPI
from sentry_logger import init

app = FastAPI()

@app.on_event("startup")
def setup_logging() -> None:
    init()
    logging.getLogger(__name__).info("Sentry SDK initialized")

Configuration Reference

Parameter	Default	Description
api_key	—	API key (sk_...). If omitted, loaded from CLI config
dsn	—	Ingest URL. Falls back to LOGSENTRY_URL env or http://localhost:8001
batch_size	50	Send logs when buffer reaches this size
flush_interval_seconds	5.0	Auto-flush buffer after this many seconds

Log Format

The SDK formats logs in the standard Python logging format. The backend parses:

2024-01-01 10:00:00,123 [INFO] [ServiceName]: Log message here

The ServiceName in brackets is used to group logs by service in the dashboard. Use Python logger names to map to service names:

logger = logging.getLogger("PaymentService")
logger.info("Payment processed")