Automate PDF Generation with n8n: HTML Templates + One HTTP Node (No Puppeteer, No Code Nodes)

<div style="font-family: system-ui; padding: 40px;">
  <h1>Invoice {{invoiceNumber}}</h1>
  <p>Customer: {{customer}}</p>
  <p>Date: {{date}}</p>
  <table>
    {{#each items}}
    <tr>
      <td>{{name}}</td>
      <td>{{price}}</td>
    </tr>
    {{/each}}
  </table>
  <p style="font-weight: bold;">Total: {{total}}</p>
</div>

/api/v1/pdf/buffer

{
  "source": {
    "type": "template",
    "data": {
      "id": "YOUR_TEMPLATE_ID",
      "context": {
        "customer": "Acme Corp",
        "invoiceNumber": "INV-001",
        "total": "€1,250.00",
        "items": [
          {"name": "Consulting", "price": "€1,000.00"},
          {"name": "Hosting", "price": "€250.00"}
        ]
      }
    }
  },
  "options": {
    "properties": {
      "size": {"width": 210, "height": 297},
      "margins": {"top": 10, "bottom": 10, "left": 10, "right": 10}
    }
  }
}

{{ $binary.data }}

{{ $("HTTP Request").item.binary.data }}
``` [[3]](https://www.width.ai/post/n8n-binary-data)

## Magic Template URL pattern: zero API body

The Magic Template URL lets you generate PDFs without any headers, JSON body, or code node [[4]](https://markupgo.com/docs/magic-template-url). The structure is:

Enable Magic URL for your template in the MarkupGo dashboard first. Then in n8n, use a single HTTP Request node with method GET and the full Magic URL as the URL field.

Pass line-item arrays with bracket syntax, since MarkupGo parses query parameters using the `qs` library with `arrayFormat: 'brackets'` [[4]](https://markupgo.com/docs/magic-template-url):

Cache reuse within 15 days serves identical URLs without consuming additional credits [[4]](https://markupgo.com/docs/magic-template-url).

The Magic URL is public — anyone with the URL can generate files and consume your credits. Add `json=true` to receive a task object response instead of the raw PDF, or `redirect=true` for a 302 redirect to the generated file. Both parameters help avoid exposing generated file URLs directly [[4]](https://markupgo.com/docs/magic-template-url).

For n8n workflows, `json=true` is usually the safer choice. You receive a task object with a `url` field, then fetch that URL in a second HTTP Request node if you need the binary file.

## Attach the PDF to email, WhatsApp, or Drive

The HTTP Request node returns binary property `data` by default when Response Format is set to File. Gmail and Send Email nodes need the binary property name to match what they expect in their attachment field.

If a downstream node expects a different binary property name, use the Edit Fields (Set) node to rename it. n8n's Edit Fields node now supports a binary data type for this purpose [[3]](https://www.width.ai/post/n8n-binary-data).

For Google Drive, set the **File** field in the Google Drive node to:

For WhatsApp Business Cloud, reference the same binary expression in the **Media** field.

The Extract From File node's **Input Binary Field** defaults to `data`, but must match the upstream node's binary property name exactly. A mismatch here produces empty files or "no data" errors [[3]](https://www.width.ai/post/n8n-binary-data).

## Retry failed renders without burning credits or creating loops

Enable **Retry On Fail** in the HTTP Request node only for HTTP status 429 and 5xx errors. Set the wait to 3 seconds with 3 attempts. This catches rate limits and transient render failures without consuming extra credits for immediate failures.

Fail immediately on 401 Unauthorized. Retrying an invalid API key creates unnecessary delay and noise. Route 401s to a separate error workflow or stop the execution.

Do not combine **Retry On Fail** with **On Error: Continue**. This combination swallows errors silently and passes undefined binary data to downstream nodes, which then fail with cryptic messages about missing attachments.

For production workflows, route all failures to a dedicated error workflow that posts to Slack or sends an email alert. Include the workflow ID, execution URL, and error message so you can diagnose without digging through execution logs.

## Set PDF expiration for sensitive documents

Add `expiresInSeconds` to the API body for auto-deletion. The range is 60 seconds to 90 days (7,776,000 seconds) [[2]](https://markupgo.com/docs/api/pdf).

```json
{
  "source": {
    "type": "template",
    "data": {
      "id": "YOUR_TEMPLATE_ID",
      "context": { /* your data */ }
    }
  },
  "expiration": {
    "expiresInSeconds": 86400
  }
}