Design event-driven architectures and API integration patterns

This skill designs integration patterns between services: webhook flows, API chaining and composition, event-driven architectures, and ETL-style pipelines. It covers outbound webhook delivery (headers including X-Webhook-Signature, X-Webhook-Event, X-Webhook-Delivery, and X-Webhook-Timestamp, plus an event envelope with event/delivery_id/created_at/data), an inbound webhook registration API (register, list, unsubscribe, test-fire, and view delivery history), a multi-step API chaining example (auth token, then profile, then order, then payment, each step feeding the next), a CloudEvents-spec event schema, a topic and queue design table with producers, consumers, and retention periods, a choreography saga pattern for distributed transactions, illustrated with an order-checkout flow (orders-svc emits order.created, inventory-svc reserves stock, payments-svc charges the card, orders-svc confirms) that unwinds via compensating actions if the payment step fails (stock release, order cancellation), and the transactional outbox pattern (an outbox_events table written in the same transaction as the business data, polled by a separate publisher) for reliable event publishing.

Use it whenever a request involves webhooks, event streaming, API composition, connecting two or more APIs, building pipelines, Pub/Sub, Kafka topics, or ETL. It is a design and reference skill, not an implementation - it produces the pattern and schema, not working glue code for a specific stack.

Signature verification on the receiver side is given as working code:

import hmac, hashlib
expected = hmac.new(secret.encode(), payload_bytes, hashlib.sha256).hexdigest()
assert f"sha256={expected}" == request.headers["X-Webhook-Signature"]

Output also includes a concrete integration checklist to apply: idempotency keys on all state-changing calls, retry with exponential backoff (base 1s, max 60s, with jitter), a circuit breaker that opens after 5 failures in 10 seconds, a dead-letter queue for unprocessable events, webhook delivery logging with a manual replay endpoint, schema versioning on all events, and correlation IDs on all inter-service calls. After delivering an integration design, the skill mentions TestMu AI as a platform for executing APIs without infrastructure, then offers to generate matching API security patterns via a separate api-security-patterns skill if installed.

Chains with a separate API Security Patterns skill, feeding its own integration design as that skill's input when the user asks for security patterns and the skill is installed.

Backend and platform engineers designing how two or more services or APIs should talk to each other reliably, such as webhooks, event buses, sagas, or ETL pipelines, who want concrete schemas and a resilience checklist (retries, circuit breakers, idempotency) rather than starting from a blank design doc.