# n8n-nodes-watsonxai

**IBM Watsonx AI Custom Node for n8n**  

This node allows you to interact with IBM Watsonx.ai for text generation using foundation models. You can generate text, summaries, or completions directly from n8n workflows.

---

## Installation

### 1. Local Installation

If you are installing locally:

```bash
npm install n8n-nodes-watsonxai

Then restart n8n. The Watsonx AI node will appear under AI in the node panel.

The Watsonx AI node allows you to generate text using IBM Watsonx foundation models.

Parameters

| Parameter | Type | Description | | --------------- | ------ | ------------------------------------------------------------------------ | | Prompt | String | Input text for the model to continue or complete. | | Model ID | String | Watsonx.ai foundation model ID (e.g., ibm/granite-13b-instruct-v2). | | Temperature | Number | Sampling temperature for text generation (0.0 = greedy, 2.0 = creative). | | Max Tokens | Number | Maximum number of tokens to generate. | | Project ID | String | IBM Cloud project ID where the model is hosted. |

Credentials

The node requires IBM Watsonx AI API credentials:

| Credential | Description | | -------------------- | ----------------------------------------------------------------------------------- | | API Key | Your IBM Cloud IAM API key. | | Watsonx Base URL | The base URL of your Watsonx.ai service, e.g., https://us-south.ml.cloud.ibm.com. |

Note: Make sure your API key has the proper permissions to access Watsonx.ai models in the selected project.

Example Workflow

1. Drag a Watsonx AI node into your workflow.
2. Select or create credentials with your API key and base URL.
3. Enter a prompt, model ID, temperature, max tokens, and project ID.
4. Connect the node to any trigger (manual or scheduled).
5. Execute the workflow to receive generated text and metadata.

Output Example:

{
  "generated_text": "Hello, this is a generated response from Watsonx AI.",
  "model_id": "ibm/granite-13b-instruct-v2",
  "stop_reason": "length",
  "token_count": 15
}

Tips

• Use short, clear prompts to get more accurate results.
• Adjust temperature for creativity: lower (0.0–0.5) = deterministic, higher (1.0–2.0) = creative.
• Keep max tokens reasonable to avoid exceeding limits and reduce latency.

License

MIT © Muhammad Muazam Arshad Email: [email protected]

---