Security and data
How the Stensul MCP server controls access, what it can and can't do in your account, and where to get Stensul's security documentation.
This page covers how access works, as the product behaves today. For data handling, model training, retention, sub-processors and certifications, use Stensul’s security documentation (below) rather than this page.
How access is controlled
- Every call acts as a signed-in person. The MCP server uses the OAuth authorization-code flow. There is no API-key or client-credentials path for MCP. See Authentication.
- Two checks on every request. The connection’s scopes and the person’s own Stensul permissions both have to allow the action. The MCP can’t do anything the person couldn’t do in the Stensul app.
- Scopes are hard limits. A connection without
emails:writecan’t create or edit emails, however it’s prompted. - Guardrails are enforced by Stensul. Locked content, palettes, fonts and character limits are checked by Stensul on every change. A change that breaks one is declined and reported; it is not reshaped or retried.
- Drafts, not sends. The MCP creates and edits drafts. It doesn’t send email, schedule it, upload it to an ESP, or route it for approval.
- Deletes need the person’s request. The only delete tool removes a module from a draft, and it asks for confirmation when more would be lost than the person named.
- Scope changes take effect on issued tokens. Adding a scope to a connection revokes its existing tokens, so everyone signs in again.
Security documentation
- Stensul Security and Trust Center
- Customers can ask their Stensul contact for the MCP security and data access FAQ.
Terms
Use of the Stensul MCP server is covered by Stensul’s product-specific terms.