How to pass an image URL or binary data to Pickaxe via n8n?

Prompt Help
1

Hey everyone,

I am trying to figure out how to structure an automated image-to-text workflow with Pickaxe and n8n, and I’d love some guidance on how to accomplish this objective.

The Goal:
I want to be able to upload a photo from my phone to a Google Drive folder. Once uploaded, an n8n workflow triggers, grabs the image link, and sends it to my Pickaxe to generate a “microblog” post based on the image. (I currently do this manually with ChatGPT, but I want to fully automate it).

The Problem:
I am currently using the Pickaxe community node in n8n. I set it up so that it sends the Google Drive link as a text message to the prompt. However, when I test it, the Pickaxe responds that it cannot access Google Drive.

I know that in the Pickaxe interface, I can manually attach and upload an image, but my goal is to have this triggered entirely by me uploading the image to a folder, completely avoiding any manual copy-and-pasting. I am using a vision-capable model for this prompt.

My Questions:

  1. Is there a way to send an image URL via the Pickaxe n8n community node so the model actually reads the image rather than just the URL text?
  2. Do I need to use a custom HTTP request in n8n to grab the file’s binary data and push it to Pickaxe? If so, how should that be formatted?
  3. Are there specific actions or configurations I need to set up within my Pickaxe to allow it to download or read an image from an external link?

Thank you so much for taking a look at this, and I look forward to hearing your suggestions!

Here is a video explaining my process..

2

Hi @enochleffingwell ,

I think the issue is that the official Pickaxe n8n node is currently sending your Google Drive link as text, not as an image input. So Pickaxe sees the URL in the prompt, but it is not really reading the image.

Google Drive links are also tricky because they usually are not direct public image URLs.

The best workaround right now is to use an HTTP Request node in n8n. Our API can accept image URLs directly.

Example:

  • POST https://api.pickaxe.co/v1/completions
  • Header: Authorization: Bearer YOUR_DEPLOYMENT_ID

Body:

{
  "message": "Write a short microblog post based on this image.",
  "imageUrls": ["DIRECT_PUBLIC_IMAGE_URL"]
}

So the flow would be:

  1. Upload image to Drive
  2. n8n grabs it
  3. n8n uploads it somewhere public with a direct image URL
  4. n8n sends that URL to Pickaxe through an HTTP request

As an alternative, you can also connect Google Drive MCP directly to the agent and let the agent work with files in Drive more natively, whether that means reading an uploaded image to write the post or generating the post and saving related assets back into Drive.

I’m also going to ask our engineers whether we can add image input support to the official n8n node.