obscurobscurIA inside your n8n workflows
The same engine that pseudonymises documents on the desktop, callable from an automation workflow — no data ever has to leave the device in the clear to be protected.
How it works, in two pieces
The node
Inside the workflow: requests protection, receives the processed text, can restore the real values later in the same workflow.
obscuria-ingress
In front of the workflow: intercepts the data BEFORE n8n receives it. Needed because n8n keeps whatever reaches it in its own execution history, even on error — a node alone, further down the workflow, always arrives one step too late.
What you need
- An active obscurIA licence on the same device where the connector runs — there is no separate one to buy or activate.
- Self-hosted n8n (Docker or a direct install). Available for Linux x86_64, Windows x64 and macOS on Apple Silicon. The macOS package is signed and notarised by Apple: no warning to bypass.
- Without a valid licence on the device, the connector refuses to process any text — not an artificial limit, the same check the desktop app runs.
Included in the annual licence
No extra cost, no separate key to configure: if you already have an active obscurIA licence on the device, the connector recognises it on its own.
Example workflows included
Text to an external AI
Protects a text, sends it to an external service, restores the real values in the response.
Definitive anonymisation
No mapping produced: for when the data must never come back, not even by mistake.
Human review
The protected text pauses for a check before continuing through the workflow.
What the connector does not change
The engine that recognises personal data, the final check before output, the security rules: they are the same as the desktop app's. There is no setting to relax them inside n8n.