n8n-nodes-apimart

Installable n8n community nodes for APIMart AI. The first release focuses on the unified asynchronous video API and supports three production patterns:

• submit and return immediately with taskId;
• query or poll GET /v1/tasks/{task_id};
• receive the terminal result through an n8n webhook trigger.

The package contains:

• APIMart AI — submit video generation and query task status;
• APIMart Callback — trigger a workflow when APIMart calls back with completed or failed.

Install

In n8n, open Settings → Community Nodes → Install and enter:

n8n-nodes-apimart

For local development:

npm install
npm run build
npm pack

Install the generated .tgz in the n8n custom/community nodes environment.

Credentials

Create an APIMart API credential with:

• API Key from https://apimart.ai/keys;
• Base URL https://api.apimart.ai/v1 (default).

The credential is sent as Authorization: Bearer ....

Workflow modes

1. Submit only

Use APIMart AI → Video → Submit Generation, select a model, and leave Response Mode as Submit Only. The node calls:

POST /v1/videos/generations

It normalizes APIMart's data[0].task_id to a top-level taskId while retaining the raw response.

2. Query / poll

Use APIMart AI → Task → Get Status with a taskId. The node calls:

GET /v1/tasks/{task_id}?language=en

It adds terminal, successful, and flattened videoUrls fields. Alternatively, select Wait Until Finished on submission to poll inside one execution.

For long videos in production, Submit Only plus Webhook is preferred: it does not hold an n8n execution open.

3. Webhook callback

1. Add APIMart Callback and activate the workflow.
2. Copy its Production URL.
3. Paste that URL into Webhook Base URL on APIMart AI → Submit Generation.

APIMart requires a base URL and appends /callback itself. The submit node detects an n8n URL already ending in /callback and removes the suffix before sending it to APIMart, so the actual request reaches the trigger URL exactly once.

Callbacks occur only for terminal states (completed or failed). APIMart may retry and may rarely deliver duplicates; deduplicate downstream by taskId / callback id.

Model parameters

All video jobs use the same endpoint, but each model has different fields and constraints. The node exposes common fields and an Extra Parameters (JSON) object. Extra parameters override common fields. This keeps the node compatible with specialized and newly released models.

The built-in catalog and full parameter matrix are documented in docs/video-models.md. Always use the exact casing documented for a model (for example MiniMax-H3 and 1080P).

Example for Seedance 2.0:

{
  "size": "adaptive",
  "generate_audio": true,
  "return_last_frame": true,
  "image_with_roles": [
    { "url": "https://example.com/start.jpg", "role": "first_frame" }
  ]
}

Webhook security

APIMart's public webhook documentation recommends signature verification, but currently does not define a standard signature header/algorithm. The callback node therefore defaults to None and offers optional Header Token and HMAC SHA-256 verification for deployments where a reverse proxy or compatible gateway adds such a header.

Official references

• Video documentation index
• Task status
• Task completion webhook
• LLM-readable documentation index

The model catalog in this release was checked against the official English pages on 2026-08-12.