Claude Webhook Integration - Async Processing with Callbacks

Webhook integration with Claude enables async AI processing for ShopMax India. Instead of making customers wait for Claude to analyze a return request or generate a product description, the system accepts the job, queues it, processes it with Claude in the background, and calls back a webhook URL when done. This pattern suits long-running tasks where synchronous HTTP timeout would be a problem.

The pattern involves three components: a receiver endpoint that accepts the job and returns 202 Accepted immediately, a background worker that calls Claude and posts the result to the callback URL, and the callback handler on the client side that receives the completed AI result. Python's threading or asyncio background tasks handle the async execution, while the callback uses a simple HTTP POST with the result payload.

The following example shows ShopMax India's async product description generator. The API accepts a product ID and callback URL, returns 202 immediately, then calls Claude in the background and posts the generated description to the webhook when ready.

It gives the following output,

# POST /api/v1/generate-description
# Body: {"job_id": "JOB-771", "product_id": "PROD-AC15",
#   "product_name": "LG 1.5 Ton 5 Star Inverter AC",
#   "specs": "5 star BEE rating, Wi-Fi enabled, auto clean, R32 refrigerant",
#   "callback_url": "http://localhost:8000/webhook/receive"}

# Immediate response (202):
{"job_id": "JOB-771", "status": "accepted",
 "message": "Description generation started. Result will be posted to callback_url."}

# Webhook callback (arrives ~2 seconds later):
Webhook received for job: JOB-771
Status: completed
Description: Beat the heat with the LG 1.5 Ton 5 Star Inverter AC - India's top ...
Callback sent to http://localhost:8000/webhook/receive for job JOB-771

In production at ShopMax India, persist job state in Redis or a database so you can query job status if the callback fails to arrive. Implement callback retry logic with exponential backoff - network hiccups are common, and a single POST failure should not lose the result. Validate the callback URL against an allowlist of known ShopMax services to prevent SSRF attacks. For bulk catalog updates during onboarding of new products, use a queue like Celery or RQ to process hundreds of description jobs concurrently without overwhelming the Claude API rate limits.