HELP - the node add attachement to a draft does not work!

I’m having a lot of trouble adding attachements to my drafts in outlook. The ADD attachement node does not work. Do you guys have any tips and tricks to solve with this problem?

(srry i typed it twice]

the specific issue here is the “add attachment” node not working for drafts in outlook.

from what i can gather, it seems like the add attachment node might be intended to work with actual attachments (e.g. files uploaded through an interface), but you’re trying to use it with a draft. that’s a good starting point.

have you tried using the “add text” or “insert link” nodes instead of the add attachment node, and then adding the attachment separately? also, what version of n8n are you running (and which platform are you on: windows, mac, linux)?

Hi @Samuel_Raymond,

Can you elaborate a bit more on what is not working?

Are you getting an error or is it not doing anything?

The sample below shows how to add attachments to a new draft or existing message. You simply need some file binary and then just reference it’s variable name

I am experiencing the same.

I even tried a super simple stupid example using this

And I am still getting the error "were unable to deserialize”

{
  "error": {
    "code": "UnableToDeserializePostBody",
    "message": "were unable to deserialize "
  }
}

I suspect that the introduction of filesystem-v2 is the cause because the Outlook API expects the base64. Although I cannot confirm that for sure.

1 Like

Hi @Lukas.Va,

When copying workflows from templates, make sure you manually upgrade the node version of outlook to the latest. Since that template is 2 years old, it is also using old nodes.

To upgrade the node version, delete and recreate it. See if this helps

Hi Wouter,

Amazing catch. The nodes were indeed old. I recreated fresh for the sake of common ground and the error is exactly the same.

The only way I got it working for now is by using the raw HTTP Request node to accommodate the company’s needs.

Would be lovely if n8n could take a look at this since Outlook is extremely common to enterprises. :slight_smile:

1 Like

Yep, I am experiencing the exact same problem as Lukas.

Also, it’s a bit trickier, but if you are responding to an email in an Outlook node and don’t want it to be sent to the person you’re replying to (so you make a draft), you can’t link an attachment in that node.

So the obvious thing to do, IMO, is to add an “Add Attachment” afterward… to add the attachment to the draft you’ve just created. But then I get the same error as Lukas. I even tried using the HTTP node to make it work, but Claude Code and I gave up after two days of trying everything.

The funny thing is, I believe it works with Gmail, but my client is using Outlook and I need it to work with Outlook.

1 Like

Perhaps I can support you. I feel your pain spending days on it.

The AI is very helpful but it requires to be babysit otherwise one cannot trust it. It has a lot of old references to n8n.

Where are you stuck at? I mean specifically node in your workflow.

The Problem

After updating from n8n 1.x to n8n 2.4.7, adding attachments to Outlook Drafts stopped working. The error message was:

{ "error": { "code": "UnableToDeserializePostBody", "message": "were unable to deserialize " } }

Root Cause

In n8n 2.x, binary data is stored on disk by default using filesystem-v2 mode instead of keeping it in memory. The Microsoft Outlook Node was not updated to handle this new storage mode. It expects Base64 data directly in memory but only finds a placeholder ("filesystem-v2").

This is a known bug:

  • GitHub Issue: #23132

  • Status: Open, assigned to the Nodes team, no fix released yet

The Solution

Change the binary data mode in the Docker configuration:

yaml

# In /root/docker-compose.yml under environment:
- N8N_DEFAULT_BINARY_DATA_MODE=default

Then restart n8n:

bash

cd /root
docker compose down
docker compose up -d

What This Setting Does

With default, binary data stays in memory (like in n8n 1.x), allowing the Outlook Node to function correctly again.

Downside: Higher RAM usage with large files. Once n8n officially fixes the bug, you can remove this line.

1 Like