Skip to main content
The Python SDK returns validated pydantic v2 models with typed, snake_cased attributes. Python 3.10+, built on httpx. 
pip install webhook-co

​
Quickstart

import os
from webhook_co import WebhookClient

client = WebhookClient(api_key=os.environ["WEBHOOK_API_KEY"])

endpoint = client.endpoints.create(name="orders-prod")
print(endpoint.ingest_url)  # returned once — capture it now

​
Pagination

List methods return a Paginator you can iterate directly: 
for endpoint in client.endpoints.list(name="prod"):
    print(endpoint.id)

# Collect everything (careful with large result sets):
failed = client.deliveries.list(status=["failed"]).collect()

# One page at a time:
page = client.endpoints.list_page(limit=50)
print(page.items, page.next_cursor)

​
Errors

Every failure is a WebhookError subclass, so you can narrow with except: 
from webhook_co import WebhookRateLimitError, WebhookNotFoundError

try:
    client.endpoints.get(endpoint_id)
except WebhookNotFoundError:
    ...  # 404
except WebhookRateLimitError as err:
    print(f"retry after {err.retry_after_s}s")
Each error carries code, status, and request_id when present.

​
Retries & idempotency

The client retries idempotent requests on transient failures with capped exponential backoff and jitter, honouring Retry-After, and never blind-retries a non-idempotent write. Replays carry an idempotency key (auto-generated if you don’t pass one): 
import uuid

client.events.replay(
    event.id,
    target={"kind": "destination", "destinationId": destination_id},
    idempotency_key=str(uuid.uuid4()),
)

​
Configuration

client = WebhookClient(api_key, max_retries=4, timeout_s=15)  # defaults: 2 retries, 30s

# As a context manager, to close the connection pool:
with WebhookClient(api_key) as client:
    client.whoami()
Full source, changelog, and issues: github.com/webhook-co/webhook.