Connect your external agent

An external agent is anything outside Wabery that stays online and writes customer replies: n8n, Make, Zapier, your app, or your own AI runtime.

It receives a webhook, decides what to say, and sends the reply through Wabery:

Wabery webhook → your logic or AI → POST /v1/messages

In onboarding, choose I already have a backend, then pick your tool.

Need Wabery to provide and host the chatbot instead? Use the hosted-agent guide.

What every setup needs

1. Create a public HTTPS endpoint in your workflow tool and paste it into Wabery. Saving it creates a signing secret, switches the project to external routing, and sends a message.received setup event.
2. Confirm the endpoint returns a 2xx response for the setup event. It has payload.test: true and synthetic channel and conversation IDs, so acknowledge it without calling the Messages API.
3. Create the reply-ready wab_live_ project key in onboarding. Keep it secret. The same key works in the shared sandbox and on connected WhatsApp channels.
4. Join the shared WhatsApp sandbox and send a customer-style message.
5. Reply by mapping payload.channel_id and payload.conversation_id from the incoming live event into this request:

POST https://api.wabery.com/v1/messages
Authorization: Bearer wab_live_...
Content-Type: application/json

{
  "channel_id": "channel_...",
  "conversation_id": "conv_...",
  "text": "Your generated reply"
}

n8n

n8n

1. Start with a Webhook node. Set HTTP Method to POST and Respond to Immediately.

2. Under Options, add Only Run If and use this expression:

   {{ $json.body.payload.test !== true }}

   Wabery’s setup event still receives 200, but n8n does not run the reply workflow with its synthetic IDs.

3. Publish the workflow and copy its Production URL, not the temporary Test URL. Paste it into Wabery and choose Save and test webhook. See n8n’s current Webhook node documentation.

4. Put your routing, data lookup, or AI nodes after the Webhook node.

5. Finish with an HTTP Request node:

   • Method: POST
   • URL: https://api.wabery.com/v1/messages
   • Authentication: Generic Credential Type
   • Generic Auth Type: Bearer Auth
   • Bearer Token: paste the raw wab_live_... key. Do not add quotes or type Bearer; n8n adds that prefix.
   • Send Body: on
   • Body Content Type: JSON
   • Specify Body: Using Fields Below

6. Add these JSON body fields. n8n puts the Wabery event under body; explicit Webhook-node references keep working after an AI or logic node changes the current item:

   Name	Value
   channel_id	{{ $('Webhook').item.json.body.payload.channel_id }}
   conversation_id	{{ $('Webhook').item.json.body.payload.conversation_id }}
   text	Map your final reply, commonly {{ $json.output }} for an AI Agent node.

   If you rename the Webhook node, use its new name in both ID expressions.

7. Publish the workflow, then send a real message through the Wabery sandbox to prove the complete reply loop.

The finished shape is Webhook → your logic / AI → HTTP Request.

Make

Make

1. Start a scenario with Webhooks → Custom webhook, add a webhook, and copy its generated URL. Make documents this as an instant trigger in its current webhook guide.
2. Paste that URL into Wabery. Choose Run once in Make before sending the Wabery test so Make discovers the event fields.
3. Put your routing, data lookup, or AI modules after the webhook.
4. Finish with HTTP v4 → Make a request. Select POST, use https://api.wabery.com/v1/messages, select application/json, and map the two IDs and reply text into the JSON body.
5. Select API key authentication and create a keychain. The key is Bearer <your Wabery key> including the prefix, placement is header, and parameter name is Authorization. This follows Make’s current API key keychain and HTTP v4 documentation.

The finished shape is Custom webhook → your scenario → HTTP v4. Turn the scenario on after the sandbox test succeeds.

Zapier

Zapier

Nota

Webhooks by Zapier is currently available on Professional, Team, and Enterprise plans, not the Free plan.

1. Start a Zap with Webhooks by Zapier and choose Catch Hook. Zapier recommends Catch Hook for parsed POST payloads and generates a unique URL. See the current Zapier webhook trigger guide.
2. Paste the Catch Hook URL into Wabery, send the Wabery test, then choose Test trigger in Zapier so the payload fields become mappable.
3. Add the actions that choose or generate the reply.
4. Add Webhooks by Zapier → POST. Send JSON to https://api.wabery.com/v1/messages, map payload.channel_id, payload.conversation_id, and the reply text, and add Authorization: Bearer <your Wabery key>.
5. Test the POST action, then publish the Zap before proving the loop in the Wabery sandbox.

The finished shape is Catch Hook → your Zap → POST action.

Your own API

Your own API

Expose a public HTTPS POST route that returns a 2xx response quickly. Read the raw request body before JSON parsing, verify X-Wabery-Signature with the project signing secret, and ignore duplicate event IDs. Then run your reply logic and call POST /v1/messages with a reply-ready project key.

Use the webhook reference for signature verification, retries, complete event fields, delivery ordering, and media handling. Use the API client guide for TypeScript, PHP, and Python examples.

The finished shape is POST webhook → your handler → Messages API.

Before connecting a client number

Do not switch to a real number until the onboarding checklist shows all six tasks complete. The final sandbox task proves both directions: Wabery delivered the inbound event, and your selected platform returned a reply through the Messages API. When it passes, connect the client’s number under Channels.

Precaución

Treat workflow webhook URLs and Wabery keys as secrets. For production, verify Wabery’s signature before using webhook data. If your workflow platform cannot verify an HMAC against the raw body safely, place a small verified gateway in front of it.