rabbitkit¶

RabbitMQ-first production toolkit for Python — safe retries, dead-letter queues, explicit acknowledgement policies, topology validation, in-process testing, and Kubernetes-ready lifecycle management.

rabbitkit CLI — dlq inspect and replay

Quick Start¶

from rabbitkit import AsyncBroker, RabbitConfig
from rabbitkit.core.config import RetryConfig

broker = AsyncBroker(RabbitConfig())

@broker.subscriber(
    queue="orders",
    retry_config=RetryConfig(max_retries=3, delays=(5, 30, 120)),
)
async def handle(body: dict) -> None:
    print(f"Order: {body}")

# Publish (simple kwargs form)
await broker.publish(routing_key="orders", body={"id": 1})

Documentation¶

Getting Started — install, first consumer, publish, retry, test
User Guide — 30 sections, beginner to advanced
API Reference — every class, method, and config field
Ack Policies — AUTO / MANUAL / NACK_ON_ERROR / ACK_FIRST
Retry & DLQ — topology, retry count tracking, error classification
Message Safety — guarantee: original never acked before retry confirmed
Ordering Guarantees — when order is and isn't preserved
Kubernetes — liveness/readiness probes, graceful shutdown, HPA
Security — TLS, HMAC signing, decompression limits, credential hygiene
Stability Policy — stable vs experimental API surface
Comparison — vs FastStream, aio-pika, Celery
Roadmap — what's shipped and what's planned
CLI Reference — topology validate/diff/apply, dlq inspect/replay, health, routes

Installation¶

pip install rabbitkit[async]    # async transport (aio-pika)
pip install rabbitkit[sync]     # sync transport (pika)
pip install rabbitkit[all]      # everything

Imports¶

# Top-level (recommended)
from rabbitkit import AsyncBroker, SyncBroker, RabbitConfig

# Alias for the async package
from rabbitkit.async_ import AsyncBroker

# Experimental features (may change between releases)
from rabbitkit.experimental import rpc, locking, dashboard