Anyone help with creating custom post/export action to be used with my GPT

Specifically, I need help with a schema that should send an image and text description to a third-party API.

The current schema looks like this:

{
	"openapi": "3.1.0",
	"info": {
		"title": "My Title",
		"description": "Post the prompt and image to mydomain.com",
		"version": "v1.0.0"
	},
	"servers": [
		{
			"url": "https://mydomain.com/wp-json/prompt-api/v1"
		}
	],
	"paths": {
		"/gpt-create-prompt/": {
			"post": {
				"description": "Post prompt data and image",
				"operationId": "PostPrompt",
				"x-openai-isConsequential": true,
				"requestBody": {
					"description": "Image and JSON data to send",
					"required": true,
					"content": {
						"multipart/form-data": {
							"schema": {
								"type": "object",
								"properties": {
									"prompt_image": {
										"type": "string",
										"format": "binary"
									},
									"prompt_data": {
										"type": "string"
									}
								}
							}
						}
					}
				},
				"responses": {
					"201": {
						"description": "Prompt created successfully"
					},
					"400": {
						"description": "Bad request"
					}
				},
				"deprecated": false
			}
		}
	},
	"components": {
		"schemas": {}
	}
}

I’m trying to figure out the same thing. currently, I believe GPT would have to type out the whole base64 (or other encoding) string to send it over an API, which would take very long and be quite inefficient.

Maybe you could have GPT send the file name (assuming the photo is snapped with a phone with some unique-ish filename) and then you could query GPhotos or something on your backend and then put it in the database? Just brainstorming.