Creating Notebooks
Deploy a new JupyterLab notebook instance from the UI or via the Admin Chat assistant.
From the UI
Section titled “From the UI”- Navigate to Jupyter Notebooks in the sidebar
- Click New Notebook
- Fill in the form:
| Field | Required | Description |
|---|---|---|
| Name | Yes | Unique identifier (lowercase, alphanumeric, hyphens). Becomes the URL slug. |
| Owner | No | Assign an owner user for object-level access control |
| Resource Preset | No | Choose a preset or configure custom resources |
- Click Create
The notebook pod starts within a few seconds. Once status shows Running, click Open to access JupyterLab.
Resource Presets
Section titled “Resource Presets”| Preset | CPU | Memory | GPU | Use Case |
|---|---|---|---|---|
| Standard | 1 core | 2Gi | — | General exploration, small datasets |
| Medium | 2 cores | 4Gi | — | Moderate data processing |
| Large | 4 cores | 8Gi | — | Large datasets, multi-model inference |
| Custom | configurable | configurable | optional | Specific requirements |
For GPU workloads, select Custom and set the GPU count. GPU nodes must be available in your cluster — see GPU Configuration.
Pre-Installed Libraries
Section titled “Pre-Installed Libraries”Each notebook instance includes:
openai— for calling Menatic AI models via the OpenAI-compatible APIrequests— HTTP clientipywidgets— interactive widgetspapermill— for running notebooks programmatically
An Example.ipynb is pre-seeded showing how to connect to the platform’s LLM API.
Pre-Configured Environment
Section titled “Pre-Configured Environment”The following environment variables are automatically injected:
| Variable | Value |
|---|---|
FTN_BASE_URL | In-cluster backend URL |
FTN_API_KEY | Project-scoped API token |
FTN_PROJECT_ID | Owning project UUID |
JUPYTER_TOKEN | Auto-generated JupyterLab auth token |
Notebook URL
Section titled “Notebook URL”Each notebook is accessible at:
https://<domain>/jupyter/<name>/Auto-Login URL
Section titled “Auto-Login URL”The platform generates a token-secured auto-login URL for each notebook. This URL includes the Jupyter token and logs you in automatically. It is displayed in the notebook detail view.
From the Admin Chat
Section titled “From the Admin Chat”You can create notebooks via the Admin Chat assistant:
Create a new Jupyter notebook named "data-analysis" for user john@example.comThe assistant uses the create_jupyter_notebook tool to deploy the instance and returns the access URL.
Naming Conventions
Section titled “Naming Conventions”Notebook names must:
- Be lowercase
- Contain only letters, numbers, and hyphens
- Start with a letter
- Be unique within the cluster