Documentation: Hugging Face Action (AI Analysis)¶

Overview¶

The Hugging Face Action is an automation node that allows you to analyze images using artificial intelligence models hosted on Hugging Face, the open AI model platform. It belongs to the AI Models node family.

In IoT and security environments, it is useful when you want to use open or specialized models (e.g., vision models such as GLM-4.5V) for camera image analysis, with the flexibility to select the model by its identifier.

When to use this action?¶

Use this action when you need to:

Analyze images from cameras with vision models hosted on Hugging Face.
Use a specific community model by identifying it by name (org/model).
Obtain descriptions (e.g., in JSON) of what is happening in an image.

Node Configuration¶

The node has two configuration tabs at the top: Form and JSON Editor.

Empty configuration of the Hugging Face node

Form View¶

1. API Key *Required¶

Select the Hugging Face credential that authenticates access (managed in a centralized and secure way).

2. Model *Required¶

The identifier of the Hugging Face model to use, in organization/model format (e.g., zai-org/GLM-4.5V).

3. Image URLs¶

The URLs of the images to analyze (one per line). Supports template expressions (e.g., {{get_snapshot_node.url}}).

4. Prompt *Required¶

The instruction or question for the model. Supports template expressions.

Configured form of the Hugging Face node

JSON Editor View¶

JSON Editor view of the Hugging Face node

JSON Structure (Input Parameters)¶

{
  "api_key": "",
  "resource": "image",
  "operation": "analyze",
  "model_id": "zai-org/GLM-4.5V",
  "image_urls": [
    "{{get_snapshot_node.url}}"
  ],
  "video_urls": [],
  "prompt": "Describe in JSON format what you see in this security camera image."
}

JSON Fields¶

Field	Type	Description
`api_key`	string	Reference to the Hugging Face credential (managed securely).
`resource`	string	Resource type (`image`).
`operation`	string	Operation (`analyze`).
`model_id`	string	Hugging Face model ID (e.g. `zai-org/GLM-4.5V`).
`image_urls`	array (string)	URLs of the images to analyze.
`video_urls`	array (string)	URLs of videos (if applicable).
`prompt`	string	The instruction/question for the model.

Output: Where the node's data comes from¶

The analysis result (text generated by the model) is available in the node's output and can be used in downstream nodes with {{node_key}}.

Usage Examples¶

Example 1: Image analysis with an open vision model¶

Use case: Analyze a camera snapshot with a Hugging Face vision model to obtain a structured description.

Model: zai-org/GLM-4.5V
Image URLs: {{get_snapshot_node.url}}
Prompt: Describe in JSON format what you see in this security camera image.

(see JSON structure above)

Validation and Errors¶

Condition	Common cause / fix
Authentication error	The Hugging Face credential is invalid or lacks permissions.
Model not found	Verify that the `model_id` (`org/model`) exists and is accessible with your token.
URLs not working	Make sure the image URLs are publicly accessible.

Best Practices¶

Use centralized credentials: Do not write the Hugging Face token in the node.
Choose a vision model: For image analysis, select a multimodal/vision model.
Request structured outputs: Ask for JSON in the prompt when you plan to process the result in downstream nodes.
Chain with snapshot capture: Typical pattern: Get snapshot → Hugging Face → condition/notification.