Skip to main content
Fintoro API supports outbound webhooks for event-driven integrations. If you need to react to changes continuously and keep an external system synchronized with minimal delay, webhooks are the right mechanism. Fintoro sends an HTTP POST request whenever one of the selected business events occurs. In Fintoro API, webhooks are designed as a thin trigger, not as a parallel snapshot API. The delivery payload always contains only the event identity, the event type, and a pointer to the resource through resource.type + resource.id. Your integration then fetches the full resource detail from the Fintoro API version it already uses.

When to use webhooks

  • when you need to react to create, update, or delete events without building change detection around frequent polling,
  • when you want to keep a local cache or downstream system in sync only after a real change,
  • when you want to separate event reception from the later fetch of the resource detail.
The recommended model is: receive the webhook, verify the signature, store webhook-id for deduplication, and only then fetch the resource detail from Fintoro API. Leave polling only as a backfill or reconciliation fallback, not as the primary trigger for change handling.

Subscription management

Manage webhook subscriptions directly through the Fintoro API. The plaintext secret is only returned when the subscription is created or when you rotate it manually. For the exact CRUD contract, request schemas, and response payloads, use the Webhook API reference. In practice:
  • you must store plainTextSecret securely from the create or rotate response,
  • changing url or subscribedEvents does not rotate the secret automatically,
  • isActive = false does not remove history, it only prevents new deliveries.

Endpoint requirements

When you create or update a subscription, a syntactically valid HTTPS URL is not enough on its own. Fintoro validates the endpoint more strictly:
  • url must use https://,
  • url must not contain a username or password,
  • the hostname must resolve to a public IP address,
  • localhost, .local, private ranges, loopback, link-local, and other reserved addresses are rejected by the backend.
The same validation runs again immediately before the webhook is sent. If the endpoint no longer satisfies those checks, Fintoro does not send the request and marks the delivery as failed instead.

Delivery payload

The webhook payload is intentionally thin:

Field semantics

The payload does not include:
  • an API version,
  • an absolute or relative resource URL,
  • a full resource snapshot,
  • embedded lookups or nested business objects.
That contract stays stable even when your integration uses a specific API version or its own fetch strategy.

Delivery headers

Every webhook request includes these headers: webhook-id and webhook-timestamp are not assembled into a separate canonical string. The signature is calculated from the raw request body exactly as it was delivered.

Signature verification

Signature verification should have four layers:
  1. check that all required webhook headers are present,
  2. check that webhook-timestamp is not too old or too far in the future,
  3. compute HMAC SHA-256 over the raw request body and compare it in constant time,
  4. deduplicate by webhook-id so that retries remain safe to process.

Pseudocode

JavaScript / Node.js-like example

How to respond to a delivery

  • Return 2xx only after the request has been safely received, verified, and stored for downstream processing.
  • If the signature or timestamp is invalid, return 401 or your receiver-specific auth failure.
  • If a downstream dependency is temporarily unavailable and you want a retry, return a non-2xx status.
  • If you accept the request and only enqueue internal work, 204 No Content is perfectly fine.
Fintoro retries failed deliveries. That means your receiver should be:
  • idempotent,
  • resilient to duplicates,
  • able to return 2xx for an already processed webhook-id.

Retry mechanism

Fintoro treats a delivery as failed when:
  • the receiver returns a non-2xx HTTP status,
  • the request times out,
  • a network-level send error occurs.
  • the endpoint no longer passes the pre-send URL, DNS, and IP safety validation.
The current delivery policy is: In practice, the typical sequence looks like this: After all attempts are exhausted, the delivery is marked as finally failed. The subscription is not automatically disabled just because one or more deliveries failed. From the receiver perspective, it is important to:
  • never assume exactly-once delivery,
  • expect the same webhook-id to arrive more than once,
  • respond quickly and move heavier logic to your internal queue,
  • return non-2xx only when you actually want Fintoro to retry the delivery.

Thin payload and the follow-up detail fetch

Recommended integration flow:
  1. receive the webhook and verify the signature,
  2. deduplicate by webhook-id,
  3. determine the correct detail endpoint from type and resource,
  4. fetch the resource detail from the Fintoro API version used by your integration,
  5. run business logic only on the fetched detail.
Examples:
  • clients.updated + resource.id = 42GET /clients/42
  • invoices.created + resource.id = 301GET /invoices/301
  • warehouse-outbound-receipts.deleted + resource.id = 88 → if the detail is already gone, process the delete locally from the event itself
For delete events, assume that the detail endpoint may no longer return the resource. In that case, the webhook event itself becomes the source of truth for the delete.

Available events

Master data

CRM

Warehouses and catalog

Documents

Production recommendations

  • do not run heavy business logic directly in the receiver HTTP thread,
  • persist the raw payload and webhook-id before downstream processing,
  • rotate secrets in a controlled way and coordinate rollout with the receiver,
  • monitor retry patterns and repeated non-2xx responses as an indicator of a receiver-side incident,
  • log webhook-id, type, resource.type, resource.id, and the X-Request-Id from the later resource fetch,
  • for delete events, do not assume the detail endpoint can still return the deleted resource.