Webhook

The Webhook block sends HTTP POST requests to external webhook endpoints with automatic webhook headers and optional HMAC signing.

Webhook Block

Configuration

Webhook URL

The destination endpoint for your webhook request. Supports both static URLs and dynamic values from other blocks.

Payload

JSON data to send in the request body. Use the AI wand to generate payloads or reference workflow variables:

{
  "event": "workflow.completed",
  "data": {
    "result": "<agent.content>",
    "timestamp": "<function.result>"
  }
}

Signing Secret

Optional secret for HMAC-SHA256 payload signing. When provided, adds an X-Webhook-Signature header:

X-Webhook-Signature: t=1704067200000,v1=5d41402abc4b2a76b9719d911017c592...

To verify signatures, compute HMAC-SHA256(secret, "${timestamp}.${body}") and compare with the v1 value.

Additional Headers

Custom key-value headers to include with the request. These override any automatic headers with the same name.

Delivery Mode

Wait for response is the default. The block waits for the endpoint and exposes its response body, HTTP status, and headers to downstream blocks.

Fire and forget (best effort) validates the URL and payload, then starts delivery in the background. The workflow continues with dispatched: true. This confirms that the request was validated and scheduled, not that the endpoint received it. Invalid requests and exhausted background capacity fail the block before dispatch:

  • The URL must be an absolute http:// or https:// URL. Anything else fails with Fire-and-forget requires an absolute HTTP(S) URL.
  • Each worker keeps at most 32 background requests in flight. When that capacity is full the block fails with Background webhook capacity reached; retry later or use Wait for response.

Background delivery can be lost if the worker shuts down. Attempts have a 30-second deadline and a 64 KiB response limit. Later delivery failures are logged but do not fail or retry the completed block. Choose Wait for response when delivery results matter or downstream blocks need the response.

Automatic Headers

Every request includes these headers automatically:

HeaderDescription
Content-Typeapplication/json
X-Webhook-TimestampUnix timestamp in milliseconds
X-Delivery-IDUnique UUID for this delivery
Idempotency-KeySame as X-Delivery-ID for deduplication

Outputs

OutputTypeDescription
datajsonResponse body; only available in Wait for response mode
statusnumberHTTP status code; only available in Wait for response mode
headersobjectResponse headers; only available in Wait for response mode
dispatchedbooleanOnly available in Fire and forget mode. Confirms scheduling, not delivery

Example Use Cases

Notify external services - Send workflow results to Slack, Discord, or custom endpoints

Agent → Function (format) → Webhook (notify)

Trigger external workflows - Start processes in other systems when conditions are met

Condition (check) → Webhook (trigger) → Response

The Webhook block always uses POST. For other HTTP methods or more control, use the API block.

Common Questions

No. The Webhook block always sends POST requests. If you need GET, PUT, DELETE, or PATCH, use the API block instead, which supports all standard HTTP methods.
When you provide a signing secret, the block generates an HMAC-SHA256 signature of the payload and includes it in the X-Webhook-Signature header in the format t=timestamp,v1=signature. The receiver can verify by computing HMAC-SHA256(secret, "timestamp.body") and comparing with the v1 value.
Every webhook request automatically includes Content-Type (application/json), X-Webhook-Timestamp (Unix timestamp in milliseconds), X-Delivery-ID (unique UUID), and Idempotency-Key (same as X-Delivery-ID for deduplication).
Yes. Any custom headers you define in the Additional Headers section will override automatic headers that share the same name.
The Webhook block is purpose-built for webhook delivery: it is POST-only, automatically adds webhook-specific headers (timestamp, delivery ID, idempotency key), and supports optional HMAC signing. The API block is more general-purpose with support for all HTTP methods, query parameters, and configurable retries.

On this page