Creation and Imports

This section explains the main ways to create a diagram in DataParade:

For code scanning inputs (CLI scan and ZIP scan), current language support is TypeScript, JavaScript, Python, and Terraform (.tf / .tfvars), including common AWS, Azure (azurerm), and Kubernetes (kubernetes_*) provider resources.

Import a dataflow.json file generated by the CLI — upload the file in the create dialog, or use CLI upload to dashboard to open preview from a scan link. See Output and results for the wrapper format.
Upload a local project ZIP and let DataParade scan it.
Start from a built-in template—see Templates for each starter diagram and what it contains.

Use these paths when you want a faster starting point instead of drawing an empty canvas. After import, fill in Node Properties and review Node Types.

CLI upload from scan (recommended)

After every successful scan, the CLI auto-uploads dataflow.json and prints a Preview ready link (unless auto-upload is skipped).

With a workspace API key: dashboard link (/dashboard?importDraft=<draftId>). Open while signed in to that workspace to load Preview & Edit without choosing a file in the create dialog.
Without a workspace API key: sign-up link (/preview/cli/<claimToken>). Create an account; the scan is attached to your new workspace and opens the same Preview & Edit modal on the dashboard.

You can also run upload <file> on an existing dataflow.json. Full CLI details: Upload to dashboard.

Import `dataflow.json` (manual)

Open the create dialog.
Choose CLI import.
Upload a valid dataflow.json.
Preview and adjust extracted content.
Create the diagram.

Preview: components, flows, and AI insights

The preview lists assets, third parties, actors, and data flows with include/exclude toggles. Open a row to edit properties in a side sheet before creating the diagram.

When the scan ran with AI inference, the summary may show an agentic trace line (candidate runs and tool-call counts from metadata.aiInference.agenticTrace).

For third-party nodes that include data-flow insights in the file:

The preview maps metadata.aiInference.thirdPartyDataFlow onto node data._cliDataFlowInsights.
In the property sheet, use the Data tab to review direction (sent to / received from vendor), data categories, confidence, capabilities, and evidence (file paths and line ranges).
These insights are read-only in preview (they do not count as “edited” fields in the diff summary) and are persisted on the diagram graph when you create the diagram.

Scans without AI, or with aiThirdPartyDataFlowEnabled: false, omit that metadata block; the Data tab appears only when insights are present.

Hosted Git and ZIP scans

From the dashboard Create assessment dialog you can scan a connected GitHub or GitLab repository or upload a ZIP archive on DataParade servers (hosted scan).

Before you start, the app runs a pre-scan project estimate (repository size or ZIP file size) and checks your workspace scan quotas. If the project is too large for your remaining budget or you have no scan slots left, Create stays disabled and a notice explains why. After a scan completes successfully, usage is deducted from Workspace → Usage; failed scans do not consume a slot.

See Workspace scan quotas for limits, when counters change, and how estimates work (hosted dashboard and CLI platform AI).

Local ZIP import

Open the create dialog.
Choose scan local folder (ZIP).
Choose a .zip file — the app checks quota against the archive size before upload (see Workspace scan quotas).
Wait for scan completion.
Open preview and create the diagram.

For Git, select a repository and ref in the create dialog; the same pre-scan estimate runs after you pick a repo.

Templates

Templates are ready-made DiagramGraphJson graphs (nodes, data flows, viewport) shipped under frontend/public/templates/. In the app, pick Template in the create flow and choose one from the dropdown.

For a plain-language tour of each template and the node types it uses, open Templates.