mamitsu
Sign inTry for free
Try for free
All guides

n8n integration

Extract social content to JSON in n8n

Import the workflow, add your mamitsu Bearer credential, and run it with a public Instagram, TikTok, YouTube, or X URL. It uses n8n's built-in HTTP Request node, so there is no community node to install.

Download the n8n workflow JSON

The workflow starts manually and is inactive on import. It contains example inputs, but no API key or saved credential. Review the URL and prompt before you execute it.

Workflow
Manual Trigger
  → Edit Fields (set URL, prompt, optional schema)
  → HTTP Request (POST /api/v1/extract)

Configure once

Store your API key in an n8n credential

The Edit Fields node contains a JSON object with url, prompt, and schema. Change those values before execution. Remove the schema property when you want mamitsu to choose the output structure.

  1. If you do not have an API key yet, create one under Mamitsu → API keys.
  2. In n8n, open the imported HTTP Request node. Under Authentication, choose Generic Credential Type, then Bearer Auth, and create a credential.
  3. Paste the API key into the credential's Bearer Token field and save it. Do not prepend Bearer; n8n adds that to the Authorization header.
  4. Select the saved credential on the HTTP Request node. The workflow export does not contain a credential reference.

Keep the key in n8n credentials, not in Edit Fields, an HTTP header value, or the exported JSON.

Request destination
POST https://usemamitsu.com/api/v1/extract
Authentication: Bearer Auth credential
Body: JSON from Edit Fields

Run it manually

Edit the sample inputs before the paid request

  1. Open the Edit Fields node. Replace the example URL with a public supported post and change the prompt to the information you need.
  2. Check the mamitsu API key has available credits. Each request that reaches the model can spend credits, including an extraction that fails.
  3. Select Execute workflow in the canvas. The HTTP Request node waits up to 120 seconds, above the API's 90-second processing timeout.
  4. Inspect the HTTP Request output. A successful response contains data, source, and extraction_id.

Import into a new workflow or an empty canvas to avoid mixing the template with nodes already on another workflow.

Import into your local n8n
Workflow menu → Import from File
Choose mamitsu-n8n-social-extraction.json

Use the output

Pass the extracted object to your next node

The HTTP Request node returns the API response body directly. In later nodes, use $json.data for the extracted fields, $json.source for the provider and URL, and $json.extraction_id when tracing a request or reporting a problem.

Treat extracted values as data to review, not as verified facts. If a value will trigger an important decision or action, add a review step before that action.

Example response shape
{
  "extraction_id": "ext_example",
  "data": {
    "dish": "Penne all'arrabbiata",
    "ingredients": ["penne", "tomatoes", "garlic"]
  },
  "source": {
    "provider": "instagram",
    "url": "https://instagram.com/reel/C8pQ2mLtN4x"
  }
}

Choose the shape

Keep or remove the optional schema

The sample workflow includes a JSON Schema for a fixed recipe object. Edit the JSON in the Edit Fields node when you need different fields. To let mamitsu choose the JSON structure from your prompt, remove the schema property from that JSON.

The API rejects unsupported JSON Schema keywords. Check the API schema reference for supported keywords and size limits.

Before production

Estimate credits and use supported public URLs

Each extraction costs at least 1 credit: 1 credit per started minute of video, plus 1 credit per started group of three photos. For YouTube, the first five minutes cost 1 credit per started minute; each started five-minute block after that costs 1 credit. The full cost is charged when the model is reached, even if extraction fails. Failures before the model are free.

Use a public HTTPS URL from Instagram, TikTok, YouTube, or X. Private or removed posts, unsupported URLs, oversized media, and provider duration limits can prevent extraction. See the API reference for the current provider limits and credit details.

The template does not retry automatically. Check the error code and extraction_id before deciding whether to retry.

Troubleshooting

Read the API error in the node output

The HTTP Request node is configured to return JSON error responses as output, including non-2xx statuses. Check $json.error.code and $json.error.message. Common causes include unauthorized (check the selected credential), insufficient_credits, unsupported_url, url_source_inaccessible, and invalid_schema.

Include extraction_id when reporting a server-side problem. For the full error list, use the API error codes.

n8n's import and export guide explains workflow JSON imports. Its HTTP Request credentials guide covers Bearer Auth setup.